“As of October 29, 2026, no federal AI statute of general application exists, and the December 2025 preemption order has not displaced a single state AI law. The one preemption ruling against a California AI law came under Section 230, as applied to two platforms. The one AI-specific statute a court has blocked in 2026 is enjoined pending appeal in a First Amendment challenge. The live exposure for a California AI company is California's own stack and the bills signed in September 2026. It also faces a twelve-state wave of chatbot laws and enforcement under ordinary consumer-protection and tort law. In California, "the AI did it" is not a defense.”
On September 30, 2026, Governor Newsom signed the last of the AI bills from California’s 2026 session. He signed thirteen that day and twenty-seven over the month, by this tracker’s count, and vetoed four more.1 The day before, the White House had ordered the federal executive branch to say “Super Intelligence” instead of “Artificial Intelligence.” The same September 30, Newsom ordered California’s agencies to keep saying “Artificial Intelligence” and “AI.”2 Four months earlier, Colorado had repealed the AI Act that compliance calendars across the country were built around, effective January 1, 2027. A narrower law takes its place the same day.3 If your company’s AI compliance plan was written in 2025, it is aimed at the wrong laws.
This tracker is current to October 29, 2026. The short version: the comprehensive frameworks shrank or stalled, the targeted statutes multiplied, and the January 1, 2027 cluster of effective dates is the one to calendar now.
Key Takeaways
- California’s live AI obligations are targeted, not comprehensive. The CCPA’s automated decisionmaking rules, the companion-chatbot law, the AI Transparency Act, the Civil Rights Council’s employment rules, and AB 316’s bar on blaming the AI are in force. SB 53’s frontier-model duties reach only developers training above 10^26 operations.4
- September 2026 added a second layer, mostly due in 2027. Twenty-seven AI bills were chaptered between September 9 and September 30. Among them are SB 947’s bar on disciplining or firing by an automated decision system alone and an AI-layoff notice rule. Others add a child-safety chatbot law, a bar on representing a customer-service chatbot as human, and an AI-auditor registry the Governor calls part of a first-in-the-nation framework. Most take effect January 1, 2027; SB 1000 and AB 502 took effect on signature.15
- Colorado’s AI Act is being replaced before it is enforced. SB 26-189, signed May 14, 2026, repeals and reenacts the law effective January 1, 2027 as a notice-and-rights law that only the Attorney General enforces. A stipulated order in xAI’s suit bars enforcement of both the 2024 duties and their replacement for conduct through 14 days after a preliminary-injunction ruling.367
- There is no federal AI law, and the federal preemption campaign has not displaced a state AI law. The December 2025 preemption order produced a task force, a proposed FTC statement, and a legislative recommendation. The Commerce evaluation of “onerous” state laws that section 4 directed had not appeared in the Federal Register as of October 29, 2026. The one preemption ruling against a California AI law came under Section 230, as applied to two platforms. The only AI statute a court blocked in 2026 is enjoined pending appeal in a First Amendment challenge.8910
- Enforcement runs under ordinary law, and the AI is treated as a tool. Ten state attorneys general have sued OpenAI. The Florida, Nebraska, and Texas announcements rest the suits on consumer-protection, deceptive-practices, and design-code law, not AI statutes. In August the Ninth Circuit held that Perplexity’s shopping agent “is a tool, not a person for statutory purposes” under the Computer Fraud and Abuse Act. The holding is limited to who “accesses” a computer.1112
Which AI laws are in force in California right now?
The table below tracks nine California AI measures in force as of October 29, 2026. They are the ones that most often reach an AI company’s products, hiring, or litigation; the selection is this tracker’s, and the election-deepfake statutes are covered after the table. Which of the nine reach a given company turns on what its product does. Each row gives who the measure covers, what it requires, when it became operative, who enforces it, and whether a private plaintiff can sue.
The frontier-model law gets the headlines and reaches the fewest companies. SB 53, the Transparency in Frontier Artificial Intelligence Act, took effect January 1, 2026. Its heaviest duties fall on developers above $500 million in annual revenue, and the Attorney General may seek civil penalties up to $1 million per violation.4 Everything else in the table reaches ordinary products.
California AI laws and regulations in force, October 2026
| Measure | Who it covers | Core duty | Operative | Enforcer and penalty | Private suit |
|---|---|---|---|---|---|
| SB 53, Transparency in Frontier AI Act (2025) | Developers training above 10^26 operations; heavier duties above $500M revenue | Publish a frontier AI framework; transparency report at deployment; report critical safety incidents to Cal OES; whistleblower protections | Jan. 1, 2026 | Attorney General; up to $1M per violation | No under the TFAIA (Attorney General only); covered employees may sue for whistleblower retaliation (Lab. Code ch. 5.1) |
| AB 2013, generative AI training-data transparency (2024) | Developers of generative AI systems made available to Californians | Post documentation of training datasets on the developer’s website | Jan. 1, 2026 | None stated in the statute | None stated |
| SB 942, California AI Transparency Act, as amended by AB 853 (2025) and SB 1000 (2026) | Every “covered provider” of a publicly accessible generative AI system in California, with no user threshold since SB 1000; large online platforms above two million monthly users under Bus. & Prof. Code § 22757.3.1, as amended by AB 2713, from Jan. 1, 2027 | Latent provenance disclosures in AI-generated or AI-altered image, video, and audio content; a free disclosure verification tool; platforms may not strip provenance data | Aug. 2, 2026, with the SB 1000 amendments effective Sept. 30, 2026 | Attorney General, city attorneys; $5,000 per violation per day | No |
| SB 243, companion chatbots (2025) | Operators of companion chatbots | Disclose the AI, a published protocol for suicidal ideation and self-harm, and, for a known minor, the AI disclosure, three-hour break reminders, and sexual-content measures of § 22602(c), which SB 1119 strikes effective Jan. 1, 2027 while its own child duties begin July 1, 2027; annual reporting from July 1, 2027 | Jan. 1, 2026; the minor measures end Dec. 31, 2026 | Private right of action for injured users; injunctive relief and damages | Yes |
| AB 316, no autonomous-AI defense (2025) | Any defendant that developed, modified, or used an AI system that caused harm | Bars the defense that the AI acted autonomously | Jan. 1, 2026 | Courts, in the underlying action | n/a |
| AB 489, health-care titles (2025) | AI systems in health care | Bars AI from implying it holds a health-care license | Jan. 1, 2026 | Licensing boards | No |
| CCPA regulations on automated decisionmaking technology, risk assessments, and cybersecurity audits (Cal. Code Regs. tit. 11, §§ 7120 to 7222) | CCPA-covered businesses using ADMT to make significant decisions; risk assessments before the processing activities § 7150(b) lists; cybersecurity audits for businesses whose processing presents significant risk to consumers’ security under § 7120(b) | Pre-use notice, opt-out (subject to exceptions), and access for ADMT; risk assessments, with submissions attested by a member of executive management; annual independent cybersecurity audits | Jan. 1, 2026; ADMT compliance by Jan. 1, 2027 for ADMT already in use (§ 7200(b)); risk assessments for pre-2026 processing by Dec. 31, 2027 and first submissions by Apr. 1, 2028; first audit reports due Apr. 1, 2028, 2029, or 2030 by revenue tier (§ 7121(a)) | California Privacy Protection Agency, administrative fines up to $2,500 per violation or $7,500 per intentional violation or violation involving known minors under 16, as adjusted; Attorney General | CCPA data-breach actions only |
| Civil Rights Council regulations on automated-decision systems under FEHA | Employers and their agents using ADS in employment decisions | Anti-discrimination duties applied to ADS; four-year retention of ADS records | Oct. 1, 2025 | Civil Rights Department; FEHA remedies | Yes, under FEHA |
| SB 1001, bot disclosure (2018) | Bots used to sell or to influence a vote | Disclose the bot | July 1, 2019 | No enforcement provision; duties cumulative with other law (§ 17942(a)) | None created by the chapter |
Two California election-deepfake statutes have been blocked in court, each only as to the parties that challenged them. In August 2025, AB 2655, the election-deepfake platform law, was held preempted by Section 230 of the Communications Decency Act as applied to X and Rumble. AB 2839 was permanently enjoined as to the plaintiffs the same month. The state’s appeal of the AB 2655 ruling is briefed, with no argument date on the Ninth Circuit’s calendar as of October 29, 2026. The Legislature rewrote AB 2839’s section by an urgency statute, AB 502, effective September 19, 2026.9 AB 2013, the training-data law, survived its challenge at the preliminary-injunction stage on March 4, 2026. The court found xAI had shown “a distinct possibility” but not a likelihood of success. The Ninth Circuit hears argument on November 18, 2026.13
If you are a CCPA-covered business and your product makes or substantially replaces a human decision on a Californian’s job, credit, housing, education, or health care—ADMT compliance is due January 1, 2027. If you use AI in hiring or discipline, the employment rules already apply.
What AI bills did California sign in September 2026?
Between September 9 and September 30, 2026, the Governor signed twenty-seven AI bills by this tracker’s count and vetoed four. The Legislature’s chaptered record, not the press coverage, is the source for what each one does.1 The bill-by-bill guide gives all twenty-seven with dates and penalties. Three themes run through the batch: the workplace, children, and who gets to audit AI.
The workplace bills matter to every employer with California staff. SB 947 provides that “An employer shall not rely solely on an ADS when making a disciplinary or termination decision.” It sets a $500 civil penalty per violation, enforced by the Labor Commissioner and public prosecutors from July 1, 2027. SB 951 amends Cal-WARN, California’s mass-layoff notice law, which reaches sites that employ 75 or more people. A notice of a layoff “caused in whole or in substantial part by an AI system or other automated technology” must now say so. AB 1331 and AB 1883 bar bathroom surveillance and AI tools that infer employees’ emotional states or collect neural data.14 The workplace guide lays these beside the FEHA and CCPA rules by use. The child-safety bills build on SB 243. SB 1119, “Adam’s Law,” requires a companion-chatbot operator that allows child users to “Perform and document a comprehensive risk assessment” before release, from July 1, 2027. Operators above $500 million in revenue owe independent child-safety audits by January 1, 2029, and smaller operators from 2032. Civil penalties run up to $5,000 per affected child for a negligent violation and $15,000 for an intentional one. SB 867 bans toys with companion chatbots from January 1, 2027 until a January 1, 2031 sunset.15 The audit bills build what the Governor’s September 18 order calls a “first-in-the-nation framework,” on agency deadlines rather than company ones. SB 813 directs the Government Operations Agency to set criteria for independent verification organizations. AB 1405 provides that “Beginning January 1, 2029, a person shall not offer, sell, or conduct a covered AI audit unless the person is registered with the agency.” The executive order moves the agency’s deadlines to May 1 and December 1, 2027. It also orders recommendations by November 16, 2026 on “Requiring the creation of a ‘kill switch’ for frontier models.”5 Using a verification organization stays voluntary; registering as an auditor does not.
Two of the quieter bills reach the most readers. AB 1609 provides that “A large private business shall not represent that a customer service chatbot is a human.” It applies to businesses above $500 million in revenue, with penalties of $5,000 for a first violation and $10,000 after. SB 574 provides that “An attorney shall not delegate the practice of law to generative artificial intelligence.” From January 1, 2027, an attorney who uses it must “Disclose the use of generative artificial intelligence to the court for all documents submitted to the court.” That reaches every California litigator who uses generative AI, this firm included.14
The September 2026 California AI signings
| Bill | Chapter | Signed | What it does | Operative |
|---|---|---|---|---|
| AB 1405, AI auditor registry | Ch. 178 | Sept. 9 | Registry of AI auditors; registration required to conduct a covered AI audit from Jan. 1, 2029 | Jan. 1, 2027 (default)16; registry by Jan. 1, 2029, moved to Dec. 1, 2027 by executive order |
| SB 813, independent verification organizations | Ch. 179 | Sept. 9 | Criteria for voluntary independent verification organizations | Jan. 1, 2027 (default); agency criteria by Jan. 1, 2028, moved to May 1, 2027 |
| SB 1119, Adam’s Law | Ch. 190 | Sept. 10 | For operators that allow child users: risk assessments before release, child-safety audits, parental notice; $5,000 or $15,000 per affected child; limited private right of action | Core duties July 1, 2027; audits by Jan. 1, 2029 |
| SB 867, chatbot toys | Ch. 189 | Sept. 10 | Ban on toys that include a companion chatbot | Jan. 1, 2027 to Jan. 1, 2031 sunset |
| AB 1159, student data | Ch. 182 | Sept. 10 | No training AI on student data; $500 per plaintiff per violation | Jan. 1, 2027; higher education July 1, 2027 |
| SB 1276, AI-generated abuse material | Ch. 192 | Sept. 10 | Adds downloading, streaming, and accessing to Penal Code 311.3’s child-exploitation crime, and AI-generated depictions to the child-abuse reporting law’s definition of sexual exploitation | Jan. 1, 2027 (default) |
| AB 1946, abuse-material reports | Ch. 185 | Sept. 10 | Social media platforms’ required reporting tool for child sexual abuse material reaches an AI-made “digital forgery”; human review absent a hash match unless the material is already blocked; block within 48 hours; up to $250,000 per day | Jan. 1, 2027 (default) |
| SB 1050, synthetic performers | Ch. 246 | Sept. 16 | Disclosure when an advertisement prominently includes a synthetic performer; UCL and FAL enforcement | Jan. 1, 2027 (default) |
| AB 502, election deepfakes | Ch. 333 | Sept. 19 | Urgency rewrite of the enjoined AB 2839 section; 30-day pre-election window; private action | Sept. 19, 2026 |
| AB 686 and AB 2025, deceptive media and rental listings | Ch. 334, 572 | Sept. 19, 27 | Deceptive election media; disclosure of digitally altered rental images | Jan. 1, 2027 (default) |
| AB 2212, digitized sexual images on campus | Ch. 388 | Sept. 20 | Creating digitized sexually explicit material without written consent, or distributing it when the distributor knows or should have known the depicted person did not provide written consent to the disclosure, becomes sexual exploitation under the Equity in Higher Education Act | Definitions operative Aug. 1, 2027 |
| AB 1609, customer-service chatbots | Ch. 733 | Sept. 28 | Businesses above $500M may not represent a chatbot as human; route to a human agent; $5,000 then $10,000; no private action | Jan. 1, 2027 (default) |
| SB 947, automated decisions in discipline and termination | Ch. 859 | Sept. 30 | No sole reliance on an automated decision system to discipline or fire; $500 per violation | July 1, 2027 |
| SB 951, AI layoff notice | Ch. 860 | Sept. 30 | Cal-WARN notice must disclose a technology displacement; existing $500-per-day penalty and private action | Jan. 1, 2027 (default) |
| AB 1331 and AB 1883, workplace surveillance | Ch. 851, 853 | Sept. 30 | No bathroom surveillance; no AI emotion inference or neural-data collection; up to $500 per violation | Jan. 1, 2027 (default) |
| SB 1000 and AB 2713, AI Transparency Act amendments | Ch. 861, 856 | Sept. 30 | SB 1000 deletes the user threshold and the manifest-disclosure option; AB 2713 revises large platforms’ provenance duties, including the bar on knowingly stripping provenance data | Sept. 30, 2026 (SB 1000); Jan. 1, 2027 (AB 2713) |
| AB 1979 and SB 503, health-care AI | Ch. 854, 857 | Sept. 30 | Providers keep independent judgment over clinical decision support; developers mitigate bias | Jan. 1, 2027 (default) |
| SB 574, attorneys and AI | Ch. 858 | Sept. 30 | No delegating the practice of law to generative AI; disclosure to the court; personal verification of every citation | Jan. 1, 2027 (default) |
| SB 1111, digital replicas | Ch. 862 | Sept. 30 | Digital replicas added to the likeness statute and to criminal false impersonation | Jan. 1, 2027 (default) |
| SB 1159, AI is not a “person” | Ch. 863 | Sept. 30 | No knowingly using AI to falsely represent that a natural person engaged with a governmental agency; AI systems are not “person[s]” under the open-government statutes | Jan. 1, 2027 (default) |
| AB 1705, nonconsensual explicit uploads | Ch. 890 | Sept. 30 | Sites that take uploaded sexually explicit content, digitally made imagery included, must use ordinary care to keep off nonconsenting or minor subjects; the depicted individual, the greater of actual damages or up to $75,000 per violation, each full day a separate violation | Jan. 1, 2027 (default) |
| AB 2392 and AB 2504, public colleges | Ch. 855, 305 | Sept. 30, 18 | Generative AI procurement standards and a community-college pilot | Jan. 1, 2027 (default) (AB 2392); AB 2504 operative only upon appropriation, pilot by July 1, 2028 |
| Vetoed: AB 2575, AB 2656, SB 903, SB 1130 | — | Sept. 30 | Clinician override of AI, public-employer notice to unions, AI in psychotherapy, wearable recorders | — |
What stalled is as telling as what passed. The automated-decision-systems bill, AB 1018, reached third reading and never received a final vote. The broad children’s chatbot bill, AB 2023, went to the inactive file on August 30. The algorithmic-pricing bills, SB 295 and SB 384, had stalled in 2025. None became law, so California’s ADMT obligations still come from the Privacy Protection Agency’s regulations, adopted in July 2025, rather than from a statute.117 The veto messages are short: SB 903, the psychotherapy bill, was “overly broad,” and AB 2656 “would impose redundant administrative layers on public agencies.”1
If you employ people in California—SB 947’s discipline rule reaches you on July 1, 2027, AI label or not, because it defines an automated decision system by what it does. If you run a site with 75 or more employees, SB 951’s layoff-notice rule reaches you on January 1, 2027.
What happened to Colorado’s AI Act?
Colorado replaced its 2024 AI Act with a narrower law before the first enforcement could begin. On May 14, 2026, Governor Polis signed SB 26-189. It repeals Part 17 of the Colorado Consumer Protection Act, the home of SB 24-205, effective January 1, 2027. It re-enacts Part 17 as a law about “covered ADMT,” automated decision-making technology “used to materially influence a consequential decision.”3 A 2025 special session had delayed the original act’s duties from February 1 to June 30, 2026. Those duties are on the books from that date until the repeal takes effect. A stipulated order in xAI’s suit pauses enforcement of them, and of their replacement, as explained below.6
The litigation explains the timing. xAI sued Colorado on April 9, 2026, and the United States intervened on April 24 on an Equal Protection theory, not preemption. On April 27 a magistrate judge entered a stipulated stay. Under it, the Attorney General “shall not initiate enforcement,” including an investigation, of SB 24-205 or any replacement enacted this session. The bar covers violations through 14 days after the court rules on xAI’s preliminary-injunction motion. That motion is due within 28 days after Colorado finally adopts its rules, and Colorado had not adopted them as of October 29, 2026.73 The order is a stipulation, and the case is stayed, not decided.
Colorado’s AI Act, before and after
| SB 24-205 (2024, repealed effective Jan. 1, 2027) | SB 26-189 (2026, effective Jan. 1, 2027; enforcement paused under the xAI stipulation) | |
|---|---|---|
| Trigger | ”High-risk” AI systems making consequential decisions | ”Covered ADMT” used to materially influence a consequential decision |
| Developer duties | Reasonable care against algorithmic discrimination; documentation to deployers, including what they need for impact assessments; public website statement; disclosure of known discrimination risks to the Attorney General and deployers within 90 days | Documentation to deployers |
| Deployer duties | Reasonable care against algorithmic discrimination; impact assessments; notices | Clear notice before use; explanation of an adverse outcome within 30 days; correction and human-review rights; three-year records |
| Enforcement | Attorney General | ”The Attorney General shall enforce this part 17 through the ‘Colorado Consumer Protection Act’, this article 1.”; 60-day cure where the Attorney General deems cure possible, none required for knowing or repeated violations |
| Private right of action | None | None |
| Rulemaking | Permitted | Required by Jan. 1, 2027 (§§ 6-1-1704(4)(b), 6-1-1705(3)); proposed 4 CCR 904-6 noticed Aug. 11, 2026, hearing Oct. 26, 2026, not adopted as of Oct. 29, 20263 |
Colorado also enacted HB 26-1263 on May 29, 2026, a conversational-AI law that adds to the same Part 17. It requires operators to estimate users’ ages. From January 1, 2027, operators must disclose the AI to every user and keep a suicide and self-harm protocol, with added safeguards for known minors.18 On this article’s reading, the xAI stipulation may reach HB 26-1263 as well, because it amends the part SB 24-205 created; the order does not name it.
If you built a Colorado impact-assessment program in 2025—that duty ends December 31, 2026. Its replacement, a notice, explanation, and appeal workflow, takes effect January 1, 2027, under the same stipulated pause on enforcement.
Which other states passed AI laws in 2026?
By the National Conference of State Legislatures’ count, states enacted 280 AI bills in 2026 out of 1,539 introduced, as of early October. Only two were comprehensive AI laws, NCSL reports: Colorado’s replacement and Connecticut’s.19 The 2026 wave was narrower and more numerous than the headlines suggest. It brought a new frontier-model law in Illinois and an amended one in New York, provenance mandates that reach mid-size generative-AI companies, and companion-chatbot statutes in ten states.
Connecticut’s Public Act 26-15, signed May 27, 2026, protects frontier-developer whistleblowers and requires companion-chatbot crisis protocols. It also adds a disclosure duty for employment AI that interacts with applicants and a provenance mandate for generative-AI providers with more than one million monthly users. Its employment-AI disclosure and notice sections (§§ 8 to 11) are “enforced solely by the Attorney General.”20 Illinois SB 315, signed July 6, 2026, requires a large frontier developer to file a disclosure statement from January 1, 2027. From January 1, 2028, it must publish a frontier AI framework. It must also retain a third party each year for an independent audit, starting then or 90 days after first qualifying, whichever is later. The Governor’s office calls that audit requirement the first in the nation.21 New York’s RAISE Act, signed December 19, 2025, was amended on March 27, 2026 to take effect January 1, 2027. It requires a frontier AI framework and 72-hour critical-incident reporting, with penalties of up to $1 million for a first violation and $3 million thereafter.22 Texas’s H.B. 149, codified in chapter 552 of the Business and Commerce Code, has been in force since January 1, 2026. The Attorney General has exclusive enforcement authority, with one exception, and must keep an online mechanism for consumer complaints. A state agency may sanction a person it licenses, registers, or certifies, including by a penalty of up to $100,000. It may do so only after a violation is found under Section 552.105 and the Attorney General recommends more enforcement.23 Utah’s HB 276 imposes provenance and detection duties from January 1, 2027 on generative-AI providers with more than one million monthly users and on large platforms.24 Washington’s HB 1170 requires generative-AI providers with more than one million monthly users to embed provenance data in AI-generated image, video, and audio content from February 1, 2027. Only the Attorney General enforces it.24 The table below selects the 2026 enactments with the widest private reach; the chatbot count after it names the chatbot statutes the table leaves out.
The 2026 state enactments that reach private companies
| State and measure | Signed | What it does | Operative | Enforcer | Private suit |
|---|---|---|---|---|---|
| Colorado SB 26-189 | May 14, 2026 | ADMT notice, explanation, and appeal rights | Jan. 1, 2027; enforcement paused under the xAI stipulation7 | Attorney General; 60-day cure where cure is possible | No |
| Connecticut PA 26-15 | May 27, 2026 | Frontier whistleblowers, AI companions, employment-AI disclosure, provenance | Phased from Oct. 1, 2026; employment Oct. 1, 2027 | Attorney General | No |
| Illinois SB 315 | July 6, 2026 | Frontier safety disclosures, incident reports, independent audits | Disclosure statement Jan. 1, 2027; frontier framework and audits from Jan. 1, 2028 | Attorney General (civil penalties, § 25); Illinois Emergency Management Agency (disclosure filings, $1,000 per day, § 18(e)) | No |
| New York S8828 (RAISE amendment) | Mar. 27, 2026 | Frontier framework, 72-hour incident reports, DFS oversight | Jan. 1, 2027 | Attorney General; $1M then $3M | No |
| Utah HB 276 | Mar. 24, 2026 | Provenance data and detection; intimate-image consent | Jan. 1, 2027 | Division of Consumer Protection | Yes, under ch. 72b |
| Washington HB 1170 | Mar. 24, 2026 | AI content provenance above one million users | Feb. 1, 2027 | Attorney General | No |
| Oregon SB 1546 | Mar. 31, 2026 | AI companion notice, crisis protocol, minor safeguards | Jan. 1, 2027 | Private action: greater of actual damages or $1,000 per violation, plus fees | Yes |
| Washington HB 2225 | Mar. 24, 2026 | Companion-chatbot notice every three hours; hourly for minors | Jan. 1, 2027 | Consumer Protection Act | Under the CPA |
| Idaho S1297 | Mar. 31, 2026 | AI disclosure, crisis protocol, and minor safeguards (reward limits, parental tools) | July 1, 2027 | Attorney General | No |
| Georgia SB 540 (Act 518) | May 11, 2026 | AI companion chatbots: disclosure, crisis protocol, no romantic role-play with minors | July 1, 2027 | Attorney General; up to $10,000 per knowing violation | No |
| Tennessee SB 1580 | Apr. 1, 2026 | AI may not be represented as a qualified mental-health professional | July 1, 2026 | Consumer Protection Act; $5,000 civil penalty per violation | Under the TCPA |
| Hawaii Act 248 | July 14, 2026 | AI-companion disclosure and safety | July 14, 2026 (reports from Jan. 1, 2028) | UDAP, HRS § 480-2 | No |
| Hawaii Act 247 | July 14, 2026 | Deepfake digital imitations, not a chatbot law | July 14, 2026 | Individual suit, AG equitable relief | Yes |
Which states have companion chatbot laws depends on the definition, which is why the trackers disagree. This tracker counts a statute when it covers a consumer-facing AI that simulates human conversation or a relationship. The statute must also require both an AI-identity disclosure and a suicide or self-harm crisis protocol. Ten states enacted one in 2026: Washington, Oregon, Idaho, Georgia, Colorado, Connecticut, Hawaii, Nebraska, Iowa, and Rhode Island. With California and New York, that makes twelve. Eight regulate a companion by name, and four (Idaho, Nebraska, Iowa, Colorado) regulate any public-facing AI that primarily simulates human conversation, subject to exclusions.2518 The state-by-state chatbot guide gives each statute’s duties and dates. Five other states’ chatbot statutes sit beside them rather than among them. Each regulates on one axis and pairs no AI-identity disclosure with a crisis protocol. Maine bars using a chatbot “to engage in trade and commerce with a consumer” in a way that may mislead a reasonable consumer into believing it is human. The bar does not apply if the consumer is clearly and conspicuously told that “the consumer is not engaging with a human being.” Utah requires a mental-health chatbot to disclose that it is AI. Nevada bars an AI provider from representing that its system “is capable of providing professional mental or behavioral health care.” Illinois bars therapy or psychotherapy services unless “conducted by an individual who is a licensed professional.” Tennessee bars representing an AI system as a qualified mental-health professional.26 The design choice that matters most is enforcement. Oregon gives an individual “[t]he greater of the individual’s actual damages or statutory damages of $1,000 for each violation.” Washington, Tennessee, and Hawaii route violations through their consumer-protection or unfair-practices statutes, and Hawaii expressly bars a private right of action. The rest of the 2026 enactments are Attorney-General-only.27
If your product talks to users as a persona or a companion—expect a disclosure duty and a crisis protocol in all twelve states, and private plaintiffs in California, Oregon, and Washington.
Is there a federal AI law in 2026?
No. The TAKE IT DOWN Act’s definitions reach intimate images “created through the use of software, machine learning, artificial intelligence.” It requires a covered platform to remove a nonconsensual intimate image “as soon as possible, but not later than 48 hours after receiving such request.” Noncompliance is treated as a violation of an FTC rule, and the FTC began enforcing the Act on May 19, 2026.28 The FY2026 defense authorization carries AI provisions that bind the Defense Department and its contractors, including a bar on DeepSeek models in contract performance. It contains nothing that preempts state law.29
Federal AI policy therefore runs through executive orders. EO 14179 of January 23, 2025 set the deregulatory posture, and the July 2025 AI Action Plan and its three companion orders followed. EO 14365 of December 11, 2025 opened the preemption campaign discussed below. EO 14409 of June 2, 2026 directed agencies to design, within 60 days, a voluntary pre-release framework for “covered frontier models.” It states that nothing in it authorizes “a mandatory governmental licensing, preclearance, or permitting requirement.” EO 14434 of September 29, 2026 directs that the executive branch “shall use the terms ‘Super Intelligence’ and ‘SI’ in place of ‘Artificial Intelligence’ and ‘AI’” and orders a statutory definition proposal within 60 days.30 NIST’s AI center is already restyled, and the Risk Management Framework remains at version 1.0 pending revision.31
Congress moved bills without passing one. The GUARD Act would bar minors from AI companions, with civil penalties up to $250,000 per violation; it reached the Senate calendar on May 11, 2026. The NO FAKES Act of 2026 reached the calendar on June 24, and the Kids Online Safety Act was ordered reported on August 5. The House passed its kids-online-safety package 267 to 117 on June 29, and the CREATE AI Act was ordered reported 29 to 0 on June 25. The Algorithmic Accountability Act, the AI LEAD Act’s products-liability regime, the SANDBOX Act, and the AI Whistleblower Protection Act sit at referral.32
The agencies have retreated from AI-specific orders and guidance and kept enforcing fraud. The FTC set aside its own Rytr order on December 22, 2025 as an undue burden on AI innovation. In a proposed policy statement published July 7, 2026, it then treated an undisclosed suppression of accuracy in AI outputs as deception. It had not finalized that statement as of October 29, 2026, and its September 2025 orders to seven companion-chatbot companies had produced no public report.33 The SEC filed settled charges on August 26, 2026 against GenesisAI, an AI start-up, over alleged misrepresentations of its revenue projections, valuation, and customer demand. It charged a fake AI trading-bot scheme on September 29.34 The CFPB withdrew its AI adverse-action circulars on May 12, 2025, and the EEOC’s AI guidance page no longer exists.35
If you were waiting for a federal statute to tell you what to build—there is none, and the agencies that remain active are the ones that prosecute deception.
Will federal preemption override state AI laws?
Not yet, and not by executive order. Preemption of a state statute requires federal law, a statute or a valid regulation, that displaces it expressly or by implication. That is why EO 14365 itself directs a “legislative recommendation” to Congress. It is also why the FTC’s proposed statement concedes that “the FTC Act does not expressly preempt State law” before arguing for implied preemption.36 As of October 29, 2026, no state AI law has been preempted, enjoined, or struck on a preemption theory advanced by the federal government. One event could change that reading: a committee vote on the FRONTIER Act, the one live preemption bill.
Executive Order 14365, directive by directive
| Directive | Deadline | Status, October 29, 2026 |
|---|---|---|
| § 3: DOJ AI Litigation Task Force | 30 days | Established by Attorney General memorandum January 9, 2026; the only courtroom act located is the Equal Protection intervention in xAI v. Weiser; no suit against any state found7 |
| § 4: Commerce list of “onerous” state laws | 90 days | Not in the Federal Register as of October 29, 20268 |
| § 5: BEAD broadband funding conditions | 90 days | No § 5 Policy Notice among NTIA’s Federal Register documents as of October 29, 2026; § 5 does not require publication there8 |
| § 6: FCC preemptive disclosure standard | 90 days after the § 4 list | Never triggered |
| § 7: FTC policy statement on preemption | 90 days | Proposed July 7, 2026, nearly four months late; not final33 |
| § 8: legislative recommendation | — | Delivered March 20, 2026;37 the FRONTIER Act (H.R. 9925) followed on July 23 and sits at referral with no committee action recorded38 |
The White House’s own recommendation pairs broad preemption with express carve-outs. It asks Congress to displace “state AI laws that impose undue burdens” and would bar states from regulating “AI development.” It keeps states’ “traditional police powers” to enforce “laws of general applicability against AI developers and users, including particular laws to protect children, prevent fraud, and protect consumers.”37 The FRONTIER Act is narrower still. It would bar states from adopting or enforcing laws that impose new substantive obligations on AI developers in three areas: frontier-risk transparency, third-party auditing, and incident reporting. It leaves deployment-side consumer, civil-rights, and privacy law in place, along with laws protecting minors from AI harms.38 Twenty-six attorneys general, including Republicans, asked Congress on September 24, 2026 for “[a] prohibition on preemption of state laws.”39
The courts have done what the order could not, on grounds the order did not supply. The one AI-specific statute blocked in 2026 is Minnesota’s “nudification” ban. The Eighth Circuit enjoined it pending appeal on October 2 in xAI’s First Amendment challenge, after the district court had denied relief.10 The one preemption win against a California AI law came under a 1996 statute, Section 230, not under the executive order.9
If your compliance plan assumes Washington will clear the state laws before 2027—it rests on a bill with no vote and an order with no court win.
Who is actually enforcing AI law?
State attorneys general, under general laws rather than AI statutes. Florida sued OpenAI in June 2026 and moved for a temporary injunction on September 28. Nebraska sued on October 14 under its Consumer Protection Act, Uniform Deceptive Trade Practices Act, and Age-Appropriate Online Design Code Act. Its attorney general announced that seven other states, Iowa, Kansas, West Virginia, Arizona, Arkansas, Idaho, and Montana, had also sued OpenAI in their own courts. Texas sued OpenAI and its chief executive on October 23 under its Deceptive Trade Practices Act, not under the AI law it enacted in 2025.12 Kentucky sued Character Technologies in January under its consumer-protection and data-protection acts. Character Technologies resolved the Garcia wrongful-death case on January 7, 2026, then filed for Chapter 11 on October 12, 2026.40 I have found no public enforcement action anywhere in the country that rests on an AI-specific statute.
That pattern is the point of my March 2026 analysis of deployer liability: the industry calls these systems agents, and the law calls them tools. An AI system has no legal personhood. So the agency-law defenses a principal can raise, such as scope of employment or independent-contractor status, are unavailable to the company that deploys one. What remains is direct liability of the operator for the tool, and decisions from 2024 through 2026 confirm it from three directions.41
The first is the operator path. On August 4, 2026, the Ninth Circuit vacated Amazon’s injunction against Perplexity’s shopping agent under the Computer Fraud and Abuse Act. It held that “the CFAA contemplates access by a person. However advanced the Assistant currently is, it is a tool, not a person for statutory purposes,” and that “[i]t is the user who ‘accesses’ Amazon’s computers.” The court expressly reserved whether Perplexity “can avoid liability for the Assistant’s actions” in tort. The holding is about who accesses a computer, not a general shield.11 California’s AB 316 reaches the same place by statute, barring a defendant from arguing that its AI acted autonomously.42 The second is the maker path. In Garcia v. Character Technologies, the court let products-liability claims proceed against the chatbot’s maker. It held the app “is a product” for claims arising “from defects in the Character A.I. app rather than ideas or expressions within the app,” and it was “not prepared to hold that Character A.I.’s output is speech.” In July 2026 a Delaware court denied Google’s motion to dismiss an AI-defamation suit in its entirety. It described AI “tools” as “allegedly employed to effectuate the defamatory ends of their makers,” but decided the motion on “established defamation caselaw” rather than the novel AI questions.43 The third is the vendor path, and it cuts the other way in vocabulary while reaching the same result. In Mobley v. Workday, the court held an AI hiring vendor could be directly liable as the employers’ statutory “agent” under Title VII. It reasoned that nothing in the statutes “distinguishes between delegating functions to an automated agent versus a live human one.” The agent there is Workday, a company, and the holding is direct vendor liability rather than a deployer’s vicarious liability for software.44
In 2026, regulators pursued AI companies under consumer-protection, data-protection, design-code, and securities law. I have found no enforcement action that rests on an AI-specific statute. In California, “the AI did it” is not a defense, by statute. Everywhere else, plan as if it will fail too.—Chanté Eliaszadeh, Principal Attorney, Astraea Counsel APC
If you deploy an AI system that acts on customers, patients, or applicants—you operate a tool, and you answer under the laws that govern the act, AI-specific or not.
What does AI compliance cost in 2026?
The only official per-business cost estimate I have found for a U.S. AI obligation is California’s, for its CCPA rules on automated decisionmaking, risk assessments, and cybersecurity audits. The Privacy Protection Agency’s final economic statement, adopted with the rules in July 2025, puts “[t]ypical business initial costs” at “$6,058 - $63,133” and “ongoing costs” at “$19,750.”17 Small businesses face initial costs of $6,058 to $36,950 and ongoing costs of $15,831.17 The agency’s cost model states its figures in 2022 dollars.45 The ten-year total fell from $9.725 billion in the proposed rules to $4.835 billion in the final ones, and cybersecurity audits are 71 percent of it.17 In September 2024 the Department of Finance said it “generally concurs with the methodology in the SRIA” for the proposed rules. It took exceptions on the state-revenue, output, investment, and employment projections.46
What drives a company’s number within that range is which rules reach it. The agency’s October 2024 model for the proposed rules assumed an ADMT program costs 240 to 360 hours of developer time. It put a risk assessment at about 120 hours of blended compliance and legal time. It put an audit at 750 to 1,250 hours depending on revenue tier, less about 30 percent for a company already running a recognized security framework.45 For the final rules the agency cut audit hours by 25 percent and risk-assessment hours by 10 percent. It also applied the 30 percent framework reduction to every business.17 On the proposed-rule model, a typical audit ran from $39,375 for a firm under $100 million already using a framework to $362,500 for a firm above $1 billion without one.45
For every other obligation in this tracker, I have found no official or market-sourced private-sector estimate, and I will not invent one. The legislative analyses for SB 53, SB 243, SB 1119, and SB 947 price only the state’s own costs. So do Colorado’s fiscal notes for both versions of its law and Texas’s fiscal note for H.B. 149. New York’s RAISE sponsor memo lists its fiscal implications as “None.” None of the opposition arguments these analyses record gives a figure, including one that cites costs to startups.47 The nearest benchmark is the European Commission’s 2021 impact assessment for the EU AI Act. It estimated compliance for a high-risk system at “around €6 000 - €7 000” on average. The study CEPS prepared to support it put a new quality-management system at EUR 193,000 to 330,000 for a company starting from nothing.48 Bias audits under New York City’s hiring-tool law have only undated vendor list prices, from $1,500 to $5,500 for a single tool.49
If a vendor quotes you a compliance cost for a law in this tracker other than the Privacy Protection Agency’s rules—ask for the source, because I have found no official one.
What should an AI company do now?
Calendar January 1, 2027 first, because more of the measures above take effect that day than on any other. The table below is the compliance calendar as of October 29, 2026. Dates marked default follow California’s rule that a non-urgency statute takes effect on January 1 after enactment.16
AI compliance calendar, November 2026 to 2029
| Date | What happens | Who it reaches |
|---|---|---|
| Nov. 16, 2026 | Recommendations due under the September 18 executive order, including on “Requiring the creation of a ‘kill switch’ for frontier models”5 | Frontier developers; state agencies |
| Nov. 18, 2026 | Ninth Circuit argument in xAI v. Bonta on AB 2013 | Generative-AI developers with California users |
| About Nov. 28, 2026 | EO 14434’s 60-day deadline for a statutory “Super Intelligence” definition proposal2 | Everyone watching federal framework text |
| Jan. 1, 2027 | Colorado SB 26-189 (enforcement paused under the xAI stipulation) and HB 26-1263; New York RAISE Act; California CCPA ADMT compliance; Connecticut §§ 4 to 6;20 Illinois SB 315 disclosure statement; Utah HB 276; Washington HB 2225; Oregon SB 1546; Rhode Island S. 2195; California SB 951, AB 1609, AB 2713, SB 574, AB 1331, AB 1883, and the other default-dated September bills; SB 243’s minor-user duties end under SB 1119 | ADMT users; frontier developers; chatbot operators; generative-AI providers; California employers and litigators |
| Feb. 1, 2027 | Washington HB 1170 provenance | Generative-AI providers above one million users |
| May 1 and Dec. 1, 2027 | California’s agency deadlines for verification-organization criteria and the AI-auditor registry under the September 18 executive order | Prospective AI auditors |
| July 1, 2027 | California SB 947 automated-discipline rules and SB 1119 child-safety duties; AB 1159 higher-education chapter; Idaho S1297; Georgia SB 540; Nebraska LB 525 and Iowa SF 2417 chatbot duties | Employers; companion-chatbot operators, SB 1119 reaching those that allow child users |
| Oct. 1, 2027 | Connecticut employment-AI disclosure for tools deployed on or after that date20 | Employers using interactive hiring AI |
| Dec. 31, 2027 | CCPA risk assessments due for processing that predates the rules | CCPA businesses with high-risk processing |
| Jan. 1, 2028 | Illinois SB 315 frontier framework and independent audits begin | Large frontier developers |
| Apr. 1, 2028 | First CCPA risk-assessment submissions, attested by an executive, for assessments conducted in 2026 and 2027; first cybersecurity audit reports for businesses with more than $100 million in 2026 revenue | CCPA businesses with risk-assessed processing; audit-covered businesses above $100M |
| Jan. 1, 2029 | California SB 1119 child-safety audits for operators above $500M that allow child users; AB 1405 auditor registration becomes mandatory | Companion-chatbot operators; AI auditors |
Download this calendar as an .ics file for Google Calendar, Outlook, or Apple Calendar. Each event links back to this tracker for its sources.
In the next thirty days, inventory every system that makes or influences a decision about a person, and map each to the rows above. Colorado’s covered-ADMT definition turns on influence over a consequential decision. California’s ADMT definition reaches technology that replaces or substantially replaces human decisionmaking, meaning a decision made from its output without qualifying human review. Neither turns on what the vendor calls the product. Pull the Privacy Protection Agency’s regulation text and the Civil Rights Council’s employment rules, and assign an owner to each duty. If you run a companion or conversational chatbot, draft the disclosure, the crisis protocol, and the minor safeguards once, to the strictest state’s standard. The ten 2026 statutes differ in details and agree in shape.
Before January 1, 2027, build the deliverables due that day. They include the ADMT pre-use notice and opt-out and the Colorado notice-and-explanation workflow, which takes effect even while enforcement is paused. If you train at frontier scale, add the New York frontier framework and the Illinois disclosure statement. Keep the employment-ADS retention and anti-discrimination controls that already apply in California. Decide whether your processing meets the § 7120(b) significant-risk thresholds that require a CCPA cybersecurity audit. If it does, find the revenue tier that sets your first report date. Then ask whether an existing SOC 2 or ISO 27001 program can anchor the audit; the agency’s own model prices that at about a 30 percent saving.
Over the longer term, treat the audit framework as the next market. California’s auditor registry becomes mandatory for anyone conducting a covered AI audit from January 1, 2029. Add its voluntary verification organizations, Illinois’s 2028 audit mandate, and SB 1119’s child-safety audits, and there is demand for a credential that does not yet exist. A company that documents its systems to an audit-ready standard now is best placed to pass the first audit quickly. And watch the two events that could flip this tracker: the Ninth Circuit’s AB 2013 decision and any committee action on the FRONTIER Act.
Contact us to map your AI systems to the rows in this tracker and sequence the deliverables due January 1, 2027.
The founders who get this right in 2027 will not be the ones who guessed correctly about federal preemption. They will be the ones who treated every AI system as a tool they are answerable for, under the laws that already applied to the decision it touches.—Chanté Eliaszadeh, Principal Attorney, Astraea Counsel APC
Conclusion
The 2026 lesson is that AI law in the United States is arriving as targeted statutes with 2027 effective dates and as enforcement under general law. It is not arriving as the comprehensive frameworks the debate was about. Which of the rows above reach your company, and in what order, is a question of what your systems do and where your users are. It is worth an hour with counsel before the January deadlines land. This tracker is updated as the law moves; the change log below records each revision.
Related Resources
- California AI Bills Signed September 2026: All Twenty-Seven, With Dates and Penalties
- California AI Workplace Laws 2027: The Rules That Reach a Hiring, Monitoring, or Firing Tool
- AI Companion Chatbot Laws by State: Twelve Statutes, One Rule, and Every Date
- State AI Law Patchwork: 50-State Compliance Map
- Federal AI Regulation Landscape: Where Things Stand in 2026
- California Frontier AI Law (SB 53): A Compliance Guide
- The EU AI Act Deadline That Did Not Move
- Not an Agent. Not a Defense: Seven Doctrines That Already Hold AI Deployers Liable
Change log
- October 29, 2026: the FAQ and Key Takeaways now say the stipulated stay in xAI v. Weiser bars enforcement of the 2024 AI Act and its replacement, SB 26-189. The bar covers conduct through 14 days after a preliminary-injunction ruling. The stay now also appears in the Colorado and state tables and the calendar. The preemption takeaway now says the federal preemption campaign has not displaced a state AI law. It adds that the one preemption ruling against a California AI law came under Section 230, as applied to two platforms. Illinois SB 315’s dates, SB 951’s 75-person site threshold, and the calendar’s Rhode Island, Nebraska, and Iowa entries are added or corrected. The tracker is current to October 29, 2026.
- October 28, 2026, later the same day: the September count moves from twenty-four to twenty-seven. AB 1946 (abuse-material reports), AB 2212 (digitized sexual images on campus), and AB 1705 (nonconsensual explicit uploads) each reach AI-made or digitally altered images, and the September table now carries a row for each.
- October 28, 2026: the companion-chatbot count moves from “at least sixteen” to twelve under a stated rule (an AI-identity disclosure and a crisis protocol, both required), with Maine, Utah, Nevada, Illinois, and Tennessee set beside the twelve; Hawaii’s Act 247 takes its own row as a deepfake law; the SB 243 row states that SB 1119 strikes the minor measures of § 22602(c) on January 1, 2027, six months before its own child duties begin.
- October 27, 2026, later the same day: the AB 1331 and AB 1883 table row now reads “up to $500 per violation,” matching Labor Code sections 1563(d) and 1582(d)(1), which set a ceiling rather than a fixed amount.
- October 27, 2026: first publication, current to the California Legislature’s September 30, 2026 signing deadline. To see this tracker’s updates highlighted in Google AI Mode and AI Overviews, you can add astraea.law as a preferred source.
This article provides general information for educational purposes only and does not constitute legal advice. Artificial intelligence regulation is evolving rapidly. Consult qualified legal counsel for advice on your specific situation.
Notes and sources
Footnotes
-
California Legislative Information, bill history and chaptered text for the 2025—2026 Regular Session (bulk data refreshed October 25, 2026), https://downloads.leginfo.legislature.ca.gov/; Office of Governor Gavin Newsom, press releases of September 9, 10, 16, 18, 19, 20, 27, 28, and 30, 2026, https://www.gov.ca.gov/2026/09/; veto messages on A.B. 2575 PDF, A.B. 2656 PDF, S.B. 903 PDF, and S.B. 1130 PDF (Sept. 30, 2026), https://www.gov.ca.gov/wp-content/uploads/2026/09/VETO-msg-AB-2656.pdf. Chaptered texts, Stats. 2026: A.B. 686, ch. 334 PDF; A.B. 1159, ch. 182 PDF; A.B. 1979, ch. 854 PDF; A.B. 2025, ch. 572 PDF; A.B. 2392, ch. 855 PDF; A.B. 2504, ch. 305 PDF; S.B. 503, ch. 857 PDF; S.B. 1050, ch. 246 PDF; S.B. 1111, ch. 862 PDF; S.B. 1159, ch. 863 PDF; S.B. 1276, ch. 192 PDF; A.B. 1946, ch. 185 PDF; A.B. 2212, ch. 388 PDF; A.B. 1705, ch. 890 PDF. The count of twenty-seven is this tracker’s: the chaptered 2026 bills in the September 2026 signings table that regulate an AI system or AI-generated content, plus two that address AI without regulating it, AB 686 (a sunset extension of the deceptive-election-media rule) and AB 2504 (a community-college pilot reskilling creative workers displaced by generative AI), and excluding budget and trailer bills that mention AI. The companion article on the September signings states the same set of twenty-seven under its own one-sentence test, which counts AB 686 for the manipulated media its extended statute reaches. ↩ ↩2 ↩3 ↩4 ↩5
-
Exec. Order No. 14434, Inaugurating the Era of Super Intelligence, 91 Fed. Reg. 63129 (Sept. 29, 2026) PDF, https://www.federalregister.gov/citation/91-FR-63129; Cal. Exec. Order No. N-10-26 (Sept. 30, 2026) (“All agencies and departments subject to my authority shall refer to Artificial Intelligence and AI as ‘Artificial Intelligence’ and ‘AI,’ notwithstanding any rebranded or different terminology used by the federal government, unless inconsistent with the law.”) PDF, https://www.gov.ca.gov/wp-content/uploads/2026/09/SIGNED_EO-N-10-26_9.30.26.pdf. ↩ ↩2
-
S.B. 26-189, 2026 Colo. Sess. Laws ch. 131 (signed May 14, 2026), §§ 6-1-1701 to 6-1-1706, 6-1-1709, and § 5 PDF, https://leg.colorado.gov/bills/sb26-189; S.B. 24-205, 2024 Colo. Sess. Laws ch. 198 (codified at Colo. Rev. Stat. §§ 6-1-1701 to 6-1-1707, repealed and reenacted effective Jan. 1, 2027, by S.B. 26-189, §§ 1, 5(1)) PDF; Colorado Department of Law, Notice of Proposed Rulemaking, 4 CCR 904-6, Automated Decision-Making Technology & Conversational Artificial Intelligence Service Rules, Colo. Sec’y of State tracking no. 2026-00360 (notice filed Aug. 11, 2026; 49 Colo. Reg. No. 16, Aug. 25, 2026), https://www.sos.state.co.us/CCR/eDocketDetails.do?trackingNum=2026-00360. ↩ ↩2 ↩3 ↩4 ↩5
-
Transparency in Frontier Artificial Intelligence Act, S.B. 53, Stats. 2025, ch. 138 PDF (Cal. Bus. & Prof. Code §§ 22757.11 to 22757.15 PDF; Cal. Lab. Code ch. 5.1); A.B. 2013, Stats. 2024, ch. 817 PDF (Cal. Civ. Code §§ 3110 to 3111 PDF); S.B. 942, Stats. 2024, ch. 291, as amended by A.B. 853, Stats. 2025, ch. 674, and S.B. 1000, Stats. 2026, ch. 861 PDF (Cal. Bus. & Prof. Code §§ 22757 to 22757.6 PDF); A.B. 2713, Stats. 2026, ch. 856 (Cal. Bus. & Prof. Code § 22757.3.1) PDF; S.B. 243, Stats. 2025, ch. 677 (Cal. Bus. & Prof. Code §§ 22601 to 22605) PDF; A.B. 489, Stats. 2025, ch. 615 (Cal. Bus. & Prof. Code §§ 4999.8 to 4999.9) PDF; S.B. 1001, Stats. 2018, ch. 892 (Cal. Bus. & Prof. Code §§ 17940 to 17943) PDF; Cal. Code Regs. tit. 11, §§ 7001(e), (j), (ddd) PDF, 7120 PDF, 7121 PDF, 7123(e)(8) PDF, 7150(b) PDF, 7152 PDF, 7155 PDF, 7157 PDF, 7200 PDF, 7220 PDF, 7221 PDF, 7222 PDF (2026); Cal. Civ. Code §§ 1798.150(c) PDF, 1798.155(a) PDF, 1798.199.90(a) PDF; Cal. Code Regs. tit. 2, §§ 11008(a) PDF, 11008.1 PDF, 11009(f) PDF, 11013(c) PDF; Chanté Eliaszadeh, California Frontier AI Law (SB 53): A Compliance Guide for AI Companies (June 14, 2026), https://astraea.law/insights/california-frontier-ai-law-sb-53. ↩ ↩2
-
S.B. 813, Stats. 2026, ch. 179 (Cal. Gov. Code §§ 8898.1, 8898.4(a)(3)) PDF; A.B. 1405, Stats. 2026, ch. 178 (Cal. Gov. Code §§ 11549.82(a), 11549.82.5) PDF; Cal. Exec. Order No. N-9-26 (Sept. 18, 2026) PDF, https://www.gov.ca.gov/wp-content/uploads/2026/09/FINAL-N-9-26-AI-EO-9.18.26-SIGNED.pdf. ↩ ↩2 ↩3
-
S.B. 25B-004, 2025 Colo. Sess. Laws (1st Extraordinary Sess.) ch. 3 (effective Nov. 25, 2025) PDF; Legislative Council Staff, Final Fiscal Note, S.B. 25B-004 (Sept. 10, 2025) PDF, https://content.leg.colorado.gov/sites/default/files/documents/2025B/bills/fn/2025b_sb25b-004_f1.pdf. ↩ ↩2
-
X.AI LLC v. Weiser, No. 1:26-cv-01515 (D. Colo.), ECF No. 1 (complaint, Apr. 9, 2026) PDF, ECF No. 17 (complaint in intervention, Apr. 24, 2026) PDF and ECF No. 24 (minute order, Apr. 27, 2026) PDF, https://storage.courtlistener.com/recap/gov.uscourts.cod.253513/gov.uscourts.cod.253513.24.0.pdf (no later docket entry modifies the order; docket last visited Oct. 29, 2026); U.S. Dep’t of Justice, Office of the Attorney General, Memorandum for All Department of Justice Employees, Artificial Intelligence Litigation Task Force (Jan. 9, 2026) PDF, https://www.justice.gov/ag/media/1422986/dl?inline. ↩ ↩2 ↩3 ↩4
-
Exec. Order No. 14365, Ensuring a National Policy Framework for Artificial Intelligence, 90 Fed. Reg. 58499 (Dec. 11, 2025) PDF, https://www.federalregister.gov/citation/90-FR-58499; FTC, Policy Statement Concerning the Suppression of Accuracy in Artificial Intelligence Systems (proposed), 91 Fed. Reg. 41638 (July 7, 2026) PDF, https://www.federalregister.gov/citation/91-FR-41638; a Federal Register search run October 29, 2026 for Commerce Department and NTIA documents published since December 11, 2025 returned no section 4 evaluation, and NTIA’s twelve Federal Register documents in that window include no section 5 BEAD Policy Notice, which section 5 does not require to be published there. ↩ ↩2 ↩3
-
Kohls v. Bonta, No. 2:24-cv-02527 (E.D. Cal.), ECF No. 98, at 2 (Aug. 20, 2025) PDF (declaring AB 2655 “preempted by 47 U.S.C. § 230(c)(1), § 230(c)(2)(B), and § 230(e)(3), as-applied to X Corp. and Rumble”) PDF and ECF No. 101 (Aug. 29, 2025) PDF, reported at 797 F. Supp. 3d 1177, 1191 PDF (defendants “PERMANENTLY ENJOINED from enforcing AB 2839 against the named Plaintiffs”); notice of appeal, ECF No. 106 (Sept. 19, 2025), “as to 98 Order”; Babylon Bee, LLC v. Bonta, No. 25-6138 (9th Cir.), docket through Aug. 12, 2026, https://www.courtlistener.com/docket/71503636/the-babylon-bee-llc-v-bonta/; A.B. 502, Stats. 2026, ch. 333 (urgency; Elec. Code § 20012) PDF. ↩ ↩2 ↩3
-
X.AI LLC v. Ellison, No. 0:26-cv-03425, 2026 U.S. Dist. LEXIS 200330 (D. Minn. Sept. 4, 2026) PDF, ECF No. 54 PDF, https://storage.courtlistener.com/recap/gov.uscourts.mnd.235231/gov.uscourts.mnd.235231.54.0_1.pdf, injunction pending appeal granted sub nom. SpaceXAI LLC v. Ellison, No. 26-2806 (8th Cir. Oct. 2, 2026) PDF. ↩ ↩2
-
Amazon.com Servs., LLC v. Perplexity AI, Inc., 184 F.4th 1083, 1091—92 (9th Cir. 2026) PDF (slip op. at 15, 17) PDF, https://cdn.ca9.uscourts.gov/datastore/opinions/2026/08/04/26-1444.pdf; Nebraska Attorney General, press release (Oct. 14, 2026), https://ago.nebraska.gov/news/attorney-general-hilgers-announces-lawsuit-against-openai-deceiving-nebraskans-about-safety; the Florida and Texas releases are cited in the note on the ten state suits. ↩ ↩2
-
Nebraska Attorney General, press release (Oct. 14, 2026), https://ago.nebraska.gov/news/attorney-general-hilgers-announces-lawsuit-against-openai-deceiving-nebraskans-about-safety; Florida Attorney General, press releases (June 1, 2026), https://web.archive.org/web/20260601184446/https://www.myfloridalegal.com/newsrelease/attorney-general-james-uthmeier-files-first-nation-state-led-lawsuit-against-openai-ceo, and (Sept. 28, 2026), https://web.archive.org/web/20260928200437/https://www.myfloridalegal.com/newsrelease/attorney-general-james-uthmeier-files-temporary-injunction-against-openai-and-its-ceo; Texas Attorney General, press release (Oct. 23, 2026), https://www.texasattorneygeneral.gov/news/releases/attorney-general-ken-paxton-sues-openai-and-ceo-sam-altman-lying-users-and-harming-texans. ↩ ↩2
-
X.AI LLC v. Bonta, No. 2:25-cv-12295, 2026 U.S. Dist. LEXIS 48883, at *25 (C.D. Cal. Mar. 4, 2026) PDF, ECF No. 35 at 11 PDF, https://storage.courtlistener.com/recap/gov.uscourts.cacd.1000848/gov.uscourts.cacd.1000848.35.0.pdf; No. 26-1591 (9th Cir.), argument calendared Nov. 18, 2026. ↩
-
S.B. 947, Stats. 2026, ch. 859 (Cal. Lab. Code §§ 1522(b)(1), 1526.1(a), (b), (e), 1526.7) PDF; S.B. 951, Stats. 2026, ch. 860 (Cal. Lab. Code §§ 1400.5(b), 1401(d)(1)) PDF; A.B. 1331, Stats. 2026, ch. 851 (Cal. Lab. Code § 1561(a)) PDF; A.B. 1883, Stats. 2026, ch. 853 (Cal. Lab. Code §§ 1581(a), 1582(d)(1)) PDF; A.B. 1609, Stats. 2026, ch. 733 (Cal. Bus. & Prof. Code §§ 22625(g), 22626(a), 22628(b), (d)) PDF; S.B. 574, Stats. 2026, ch. 858 (Cal. Bus. & Prof. Code § 6068.1(a)(2), (a)(3)(C); Cal. Civ. Proc. Code § 128.7(b)(2)(A)) PDF; chaptered texts at https://leginfo.legislature.ca.gov/. ↩ ↩2
-
S.B. 1119, Stats. 2026, ch. 190 (Cal. Bus. & Prof. Code §§ 21810.5(j)(1)(B), 21812(a)(1), (e), 21814(a)(1)(A)(i), (c), 21816(a)(1)(A), (B)) PDF; S.B. 867, Stats. 2026, ch. 189 (Cal. Bus. & Prof. Code § 22604.5(a), (b)) PDF; Office of the Governor, press release (Sept. 10, 2026), https://www.gov.ca.gov/2026/09/. ↩
-
Cal. Const. art. IV, § 8(c)(1) (a statute enacted at a regular session goes into effect on January 1 next following a 90-day period from enactment, with stated exceptions) PDF, https://leginfo.legislature.ca.gov/faces/codes_displayText.xhtml?lawCode=CONS&division=&title=&part=&chapter=&article=IV. ↩ ↩2
-
California Privacy Protection Agency, Final Statement of Reasons and Updated Informative Digest, CCPA Updates, Cyber, Risk, ADMT, and Insurance Regulations, at 54—56 (2025) PDF, https://cppa.ca.gov/regulations/pdf/ccpa_updates_cyber_risk_admt_fsor_and_uid.pdf. ↩ ↩2 ↩3 ↩4 ↩5
-
H.B. 26-1263, 2026 Colo. Sess. Laws ch. 208 (signed May 29, 2026) PDF, §§ 6-1-1701(3.5), 6-1-1708, https://leg.colorado.gov/bill_files/116851/download. ↩ ↩2
-
National Conference of State Legislatures, As Congress Deliberates, States Lead the Way on AI Regulation (Oct. 21, 2026), https://www.ncsl.org/state-legislatures-news/details/as-congress-deliberates-states-lead-the-way-on-ai-regulation. ↩
-
Conn. Pub. Act No. 26-15 (S.B. 5) (approved May 27, 2026), §§ 2, 4 to 6, 9, 12, 15 PDF, https://www.cga.ct.gov/2026/ACT/PA/PDF/2026PA-00015-R00SB-00005-PA.PDF. ↩ ↩2 ↩3
-
Ill. Pub. Act 104-538 (S.B. 315) §§ 10(d), 15(c), 18, 25, 99 (approved July 6, 2026) PDF; Office of Governor JB Pritzker, press release on S.B. 315 (July 6, 2026), https://gov-pritzker-newsroom.prezly.com/gov-pritzker-signs-nation-leading-artificial-intelligence-safety-law. ↩
-
N.Y. Gen. Bus. Law §§ 1420(12), (16) PDF, 1421(1) PDF, 1422(1), (3)(a), (4) PDF, 1427(1) to (2) PDF (effective Jan. 1, 2027), enacted by S. 6953-B, ch. 699, L. 2025 (signed Dec. 19, 2025), as amended by S. 8828, ch. 96, L. 2026 (signed Mar. 27, 2026), https://nyassembly.gov/leg/?bn=S08828&term=2025&Summary=Y&Actions=Y. ↩
-
Tex. Bus. & Com. Code §§ 552.101 PDF, 552.102 PDF, 552.106 PDF (eff. Jan. 1, 2026) (enacted by Act of 2025, 89th Leg., R.S., ch. 1174 (H.B. 149), § 4), https://capitol.texas.gov/tlodocs/89R/billtext/html/HB00149F.htm. ↩
-
Utah Code §§ 13-72c-101 PDF, 13-72c-201 PDF, 13-72c-203, 13-72b-202 (H.B. 276, 2026 Gen. Sess.) PDF, https://le.utah.gov/Session/2026/bills/enrolled/HB0276.pdf; Wash. H.B. 1170, ch. 167, §§ 1(2), 2(1), 4(3), 9, Laws of 2026 PDF. ↩ ↩2
-
Wash. Substitute H.B. 2225, ch. 168, Laws of 2026 PDF, https://app.leg.wa.gov/billsummary?BillNumber=2225&Year=2025; Idaho S. 1297, ch. 249, 2026 Idaho Sess. Laws PDF, https://legislature.idaho.gov/sessioninfo/2026/legislation/S1297/; Ga. S.B. 540, Act 518 (approved May 11, 2026) (codified at O.C.G.A. § 39-5-6) PDF; 2026 Tenn. Pub. Acts ch. 647 (S.B. 1580) PDF; Act 247, 2026 Haw. Sess. Laws (H.B. 2137) PDF; Act 248, 2026 Haw. Sess. Laws (S.B. 3001) PDF; Neb. L.B. 525 (approved Apr. 14, 2026), https://nebraskalegislature.gov/bills/view_bill.php?DocumentID=59601&docnum=LB525&leg=109; Iowa S.F. 2417 (signed May 2, 2026), https://www.legis.iowa.gov/legislation/BillBook?ga=91&ba=SF2417; R.I. S. 2195 Sub. A, 2026 R.I. Pub. Laws ch. 376 (signed June 22, 2026), https://webserver.rilegislature.gov/PublicLaws/law26/law26376.htm; Utah H.B. 452, 2025 Gen. Sess. (Utah Code §§ 13-72a-101 to 13-72a-301; signed Mar. 25, 2025, effective May 7, 2025) PDF, https://le.utah.gov/Session/2025/bills/enrolled/HB0452.pdf. ↩
-
Me. P.L. 2025, ch. 294, § 1 (L.D. 1727, approved June 12, 2025), enacting 10 M.R.S. § 1500-Y, https://legislature.maine.gov/legis/bills/getPDF.asp?paper=HP1154&item=3&snum=132; Utah H.B. 452 (2025), Utah Code § 13-72a-101 et seq. PDF; Nev. A.B. 406 (2025), § 7, https://archive.leg.state.nv.us/Session/83rd2025/Bills/AB/AB406_EN.pdf; Ill. P.A. 104-0054 (H.B. 1806, eff. Aug. 1, 2025), Wellness and Oversight for Psychological Resources Act, synopsis at https://ilga.gov/Legislation/BillStatus?DocNum=1806&GAID=18&DocTypeID=HB&SessionID=114&GA=104; 2026 Tenn. Pub. Acts ch. 647, § 1 PDF. ↩
-
Or. S.B. 1546, 2026 Or. Laws ch. 85, § 2(1)(a) PDF, https://olis.oregonlegislature.gov/liz/2026R1/Measures/Overview/SB1546; 2026 Tenn. Pub. Acts ch. 647, § 1; Act 248, 2026 Haw. Sess. Laws, § 3. ↩
-
TAKE IT DOWN Act, Pub. L. No. 119-12, §§ 2(a), 3(a)(3), (b)(1), 139 Stat. 55, 55, 60—61 (2025) PDF, https://www.govinfo.gov/app/details/PLAW-119publ12; FTC, FTC Begins Enforcing the TAKE IT DOWN Act (May 19, 2026) PDF, https://www.ftc.gov/news-events/news/press-releases/2026/05/ftc-begins-enforcing-take-it-down-act. ↩
-
National Defense Authorization Act for Fiscal Year 2026, Pub. L. No. 119-60, §§ 1513, 1532, 139 Stat. 718, 1149—51, 1157—59 (2025) PDF, https://www.govinfo.gov/app/details/PLAW-119publ60. ↩
-
Exec. Order No. 14179, Removing Barriers to American Leadership in Artificial Intelligence, 90 Fed. Reg. 8741 (Jan. 23, 2025) PDF; Exec. Order No. 14409, Promoting Advanced Artificial Intelligence Innovation and Security, 91 Fed. Reg. 34565, § 3(b)—(c) (June 2, 2026) PDF, https://www.federalregister.gov/citation/91-FR-34565; Exec. Order No. 14434, Inaugurating the Era of Super Intelligence, 91 Fed. Reg. 63129, §§ 1, 3(b) (Sept. 29, 2026) PDF. ↩
-
NIST, AI Risk Management Framework, https://www.nist.gov/itl/ai-risk-management-framework; NIST, Center for Advancing Innovation and Standards for Super Intelligence, https://www.nist.gov/caissi. ↩
-
S. 3062, 119th Cong. (GUARD Act), reported and placed on the Senate calendar May 11, 2026, https://www.govinfo.gov/bulkdata/BILLSTATUS/119/s/BILLSTATUS-119s3062.xml; S. 4591 (NO FAKES Act of 2026), placed on the calendar June 24, 2026, https://www.govinfo.gov/bulkdata/BILLSTATUS/119/s/BILLSTATUS-119s4591.xml; S. 1748 (Kids Online Safety Act), ordered reported Aug. 5, 2026, https://www.govinfo.gov/bulkdata/BILLSTATUS/119/s/BILLSTATUS-119s1748.xml; H.R. 7757 (KIDS Act), passed the House June 29, 2026, Roll No. 228, https://www.govinfo.gov/bulkdata/BILLSTATUS/119/hr/BILLSTATUS-119hr7757.xml; H.R. 2385 (CREATE AI Act), ordered reported June 25, 2026, https://www.govinfo.gov/bulkdata/BILLSTATUS/119/hr/BILLSTATUS-119hr2385.xml; S. 2164, S. 2937, S. 2750, and S. 1792, all at referral, https://www.govinfo.gov/bulkdata/BILLSTATUS/119/s/BILLSTATUS-119s2164.xml, https://www.govinfo.gov/bulkdata/BILLSTATUS/119/s/BILLSTATUS-119s2937.xml, https://www.govinfo.gov/bulkdata/BILLSTATUS/119/s/BILLSTATUS-119s2750.xml, https://www.govinfo.gov/bulkdata/BILLSTATUS/119/s/BILLSTATUS-119s1792.xml. ↩
-
Order Reopening and Setting Aside Order, In re Rytr LLC, FTC Docket No. C-4806 (Dec. 22, 2025) PDF; FTC, Policy Statement Concerning the Suppression of Accuracy in Artificial Intelligence Systems (proposed), 91 Fed. Reg. 41638 (July 7, 2026) PDF; FTC, press release, FTC Launches Inquiry into AI Chatbots Acting as Companions (Sept. 11, 2025) (6(b) orders to Alphabet, Character Technologies, Instagram, Meta, OpenAI, Snap, and X.AI) PDF, https://www.ftc.gov/news-events/news/press-releases/2025/09/ftc-launches-inquiry-ai-chatbots-acting-companions. ↩ ↩2
-
SEC v. GenesisAI Corp., Litigation Release No. 26619 (Aug. 26, 2026) PDF; SEC v. TSAI Pro Ltd., Litigation Release No. 26654 (Sept. 29, 2026) PDF, https://www.sec.gov/enforcement-litigation/litigation-releases. ↩
-
CFPB, Interpretive Rules, Policy Statements, and Advisory Opinions; Withdrawal, 90 Fed. Reg. 20084, 20086 (May 12, 2025) PDF, https://www.federalregister.gov/citation/90-FR-20084; the EEOC’s former technical-assistance page on AI, https://www.eeoc.gov/ai, returns no page as of Oct. 29, 2026 (archived Jan. 26, 2025 at https://web.archive.org/web/20250126101228/https://www.eeoc.gov/ai). ↩
-
Exec. Order No. 14365, §§ 3—8, 90 Fed. Reg. at 58499—501 PDF; 91 Fed. Reg. at 41641 PDF. ↩
-
The White House, National Policy Framework for Artificial Intelligence: Legislative Recommendations § VII, at 4 (Mar. 20, 2026), https://www.whitehouse.gov/wp-content/uploads/2026/03/03.20.26-National-Policy-Framework-for-Artificial-Intelligence-Legislative-Recommendations.pdf. ↩ ↩2
-
H.R. 9925, 119th Cong. § 9(a)—(c) (introduced July 23, 2026) PDF, https://www.congress.gov/bill/119th-congress/house-bill/9925. ↩ ↩2
-
Office of the New York Attorney General, Attorney General James Calls on Congress to Protect Americans from Unchecked AI Development (Sept. 24, 2026), https://ag.ny.gov/press-release/2026/attorney-general-james-calls-congress-protect-americans-unchecked-ai-development. ↩
-
Garcia v. Character Technologies, Inc., No. 6:24-cv-01903 (M.D. Fla.), docket, https://www.courtlistener.com/docket/69300919/; In re Character Technologies, Inc., No. 26-11603 (Bankr. D. Del. filed Oct. 12, 2026); Kentucky Attorney General, press release (Jan. 8, 2026), https://www.kentucky.gov/Pages/Activity-stream.aspx?n=AttorneyGeneral&prId=1857. ↩
-
Chanté Eliaszadeh, Not an Agent. Not a Defense: Seven Doctrines That Already Hold AI Deployers Liable (Mar. 18, 2026), https://astraea.law/insights/ai-agent-deployer-liability. ↩
-
A.B. 316, Stats. 2025, ch. 672 PDF (Cal. Civ. Code § 1714.46 PDF). ↩
-
Garcia v. Character Technologies, Inc., 785 F. Supp. 3d 1157, 1179—80 (M.D. Fla. 2025) PDF; Starbuck v. Google LLC, C.A. No. N25C-10-211 MAA, 2026 Del. Super. LEXIS 341, slip op. at 1 (Del. Super. Ct. July 24, 2026) PDF, https://courts.delaware.gov/opinions/download.aspx?id=398660. ↩
-
Mobley v. Workday, Inc., 740 F. Supp. 3d 796, 806—07 (N.D. Cal. 2024) PDF (ECF No. 80 at 9—10) PDF, https://www.courtlistener.com/docket/66831340/. ↩
-
California Privacy Protection Agency, Standardized Regulatory Impact Assessment, ISOR Appendix A, at 49—63 (Oct. 2024) PDF, https://cppa.ca.gov/regulations/pdf/ccpa_updates_cyber_risk_admt_ins_impact.pdf. ↩ ↩2 ↩3
-
California Department of Finance, comment letter on the CPPA Standardized Regulatory Impact Assessment, at 1 (Sept. 20, 2024) PDF, https://dof.ca.gov/media/docs/forecasting/economics/major-regulations/major-regulations-table/CPPA-Finance-comment-letter-ADA.pdf. ↩
-
Senate Floor Analysis of S.B. 53 (Unfinished Business, as amended Sept. 29, 2025) PDF, https://leginfo.legislature.ca.gov/faces/billAnalysisClient.xhtml?bill_id=202520260SB53; Senate Floor Analysis of S.B. 243 (Unfinished Business, as amended Sept. 4, 2025); Assembly Appropriations Committee Analysis of S.B. 1119 (Aug. 5, 2026); Senate Floor Analysis of S.B. 947 (Unfinished Business, as amended Aug. 21, 2026); Colorado Legislative Council Staff, Final Fiscal Note, S.B. 24-205 (Aug. 12, 2024) PDF, https://content.leg.colorado.gov/sites/default/files/documents/2024A/bills/fn/2024a_sb205_f1.pdf, and Final Fiscal Note, S.B. 26-189 (Sept. 8, 2026) PDF, https://leg.colorado.gov/bill_files/117715/download; Texas Legislative Budget Board, Fiscal Note, H.B. 149 (as passed 2d House) (May 26, 2025) PDF, https://capitol.texas.gov/tlodocs/89R/fiscalnotes/pdf/HB00149F.pdf; N.Y. S. 6953-A sponsor memorandum, https://www.nysenate.gov/legislation/bills/2025/S6953/amendment/A. ↩
-
European Commission, Impact Assessment Accompanying the Proposal for a Regulation Laying Down Harmonised Rules on Artificial Intelligence, SWD(2021) 84 final, pt. 1, at 67—69, 71—72 (Apr. 21, 2021) PDF, https://artificialintelligenceact.eu/wp-content/uploads/2022/06/AIA-COM-Impact-Assessment-1-21-April.pdf; Moritz Laurer, Andrea Renda & Timothy Yeung, Clarifying the Costs for the EU’s AI Act, CEPS (Sept. 24, 2021), https://www.ceps.eu/clarifying-the-costs-for-the-eus-ai-act/. ↩
-
Vendor list prices published by Bias Audit USA, https://biasauditusa.com/pricing/, and AEDT Audits, https://aedtaudits.com/, accessed Oct. 29, 2026; neither is a market survey. ↩
