The State AI Law Map: What Takes Effect in California, New York, Colorado

On Sept. 30, 2026, Gavin Newsom signed 13 bills in a single release his office headlined as making California’s “nation-leading AI framework” stronger. “AI should expand opportunity – not come at the expense of workers and families,” the governor said. Most of those laws don’t bite yet. They switch on between now and mid-2027, alongside New York’s frontier-model law and Colorado’s rewritten AI act.

This is our running map of what takes effect when, built for two readers: the worker or consumer who wants to know what changes for them, and the company that has to comply. Effective dates come from the bill text or the state’s own legislative records. One rule of thumb for California: under the state constitution, a regular-session law goes into effect on the January 1 after a 90-day period from enactment, so anything signed in September 2026 without its own date starts Jan. 1, 2027. Urgency statutes take effect immediately.

This edition reflects laws signed through Sept. 30, 2026. We update it quarterly.

1. California SB 947, the “no robo bosses” rule: July 1, 2027

From mid-2027, a California employer can’t let software fire or discipline you on its own. The enrolled text says plainly: “An employer shall not rely solely on an ADS when making a disciplinary or termination decision.” If an automated decision system is the primary basis, a human has to corroborate the call, and the worker gets written notice and can request a description of their own data the system used. The bill was chaptered on Sept. 30 and covers public employers too. Penalty: $500 per violation, enforced by the Labor Commissioner.

It’s narrower than labour wanted. CalMatters reports the bill lost an appeals process, a private right to sue and coverage of contractors on its way through. California Labor Federation president Lorena Gonzalez still called the principle worth having: “an individual should not have the indignity of being fired or disciplined by a robot, by an algorithm, without human oversight.”

For your company: inventory every tool that scores, flags or ranks employees, and build a documented human-review step before July 2027.

2. California AB 1883, no emotion reading or brain data at work: Jan. 1, 2027

Employers may not use an AI-powered workplace surveillance tool to recognize, or make inferences or predictions about, a worker’s emotional state, or to collect neural data, under AB 1883. Neural data means information “generated by measuring the activity of an employee’s central or peripheral nervous system.” The definition of a surveillance tool is broad, from video and audio to time-tracking and geolocation. The bill carries no date of its own, so it starts Jan. 1, 2027. Narrow exemptions cover federal aerospace and defence work where the tool is needed to comply with federal law or a federal contract.

For you: that “sentiment score” on your call-centre dashboard, if it’s inferring your mood, is on borrowed time.

3. California SB 951, AI layoff notices: Jan. 1, 2027

A correction to a figure that has circulated: SB 951 does not create a 90-day notice period. The enrolled bill keeps the existing Cal-WARN clock of 60 days for mass layoffs, relocations and terminations at establishments with 75 or more employees. What changes is the content. When the cuts are substantially due to automation, the notice must say “This notice is for a technology displacement,” and list the affected jobs, the functions being automated and the category of AI or other technology replacing them. The state’s Employment Development Department has to publish a quarterly statewide summary and report to the legislature by Jan. 1, 2028.

For your company: layoff notices become public evidence of what you automated. Communications and legal need to write them together.

4. California SB 1000 and AB 2713, content provenance: now, then Jan. 1, 2027

These two amend the California AI Transparency Act, which has applied to generative AI providers since Aug. 2, 2026. SB 1000 is an urgency statute, so its changes took effect on signing: it drops the one-million-user threshold for covered providers, swaps “AI detection tool” for “disclosure verification tool” and requires hidden (latent) disclosures to say whether content was created or altered by AI. Penalty: $5,000 per violation.

AB 2713 governs large online platforms from Jan. 1, 2027. They have to surface provenance data showing content was AI-generated or substantially altered, let users inspect it, and not knowingly strip compliant provenance data where technically feasible. Platforms only have to honour data built to widely adopted standards.

For you: expect “made with AI” labels with a click-through to the underlying metadata on major platforms next year.

5. California SB 1111 and SB 1050, digital replicas and synthetic ad actors: Jan. 1, 2027

SB 1111 writes digital replicas into California’s right-of-publicity law: “voice” and “likeness” now include a computer-generated, highly realistic representation readily identifiable as a person. It also makes clear that false-impersonation crimes apply when replicas are used to deceive or defraud. A companion law, SB 1050, signed Sept. 16, requires ads that prominently feature a synthetic performer to say so, with wording such as “This performance features a synthetic performer.” Ads for films, shows and games are exempt.

6. California SB 1119, companion chatbots and kids: July 1, 2027

Signed Sept. 10 and known as Adam’s Law, SB 1119 requires chatbot operators that allow child users to run risk assessments, maintain crisis protocols with hotline referrals and ship protective defaults, including persistent memory and push notifications off and usage capped at two hours a day. Independent child-safety audits are due by Jan. 1, 2029. Operators under $500 million in annual revenue are exempt from the audit requirement until 2032.

7. New York’s RAISE Act, frontier developers: Jan. 1, 2027

New York signed the original RAISE Act in December 2025, then rewrote it. The chapter amendment, S8828, signed March 27, 2026, applies to “large frontier developers” with more than $500 million in annual revenue and takes effect Jan. 1, 2027. Covered developers must publish a frontier AI framework and pre-deployment transparency reports, report critical safety incidents to an office within the Department of Financial Services within 72 hours, and face penalties of $1 million for a first violation and up to $3 million after that. The sponsor’s memo says the rewrite aligns New York more closely with California’s SB 53.

For you: little you’ll see directly. For the handful of labs it covers, it’s a second regulator with its own deadlines.

8. Colorado SB 26-189, automated decisions: Jan. 1, 2027

Colorado has rewritten its 2024 AI Act. According to the state attorney general, SB 26-189, signed in May 2026, “repeals and reenacts” those provisions around automated decision-making technology, effective Jan. 1, 2027. The chaptered bill applies to consequential decisions made on or after that date. Deployers must give clear notice before using the technology to materially influence a decision, explain an adverse outcome within 30 days, let consumers correct inaccurate data and offer meaningful human review “to the extent commercially reasonable.” The attorney general enforces it under the Consumer Protection Act and is taking public comment on draft rules until Oct. 26, 2026.

9. Already in force: Texas, Illinois and California’s SB 53

Three laws switched on Jan. 1, 2026. Texas’s HB 149 set rules for AI systems in the state with civil penalties. Illinois’s HB 3773 amended its Human Rights Act so employers can’t use AI in ways that discriminate against protected classes, or use zip codes as a proxy, and must tell workers when AI is in use. California’s SB 53, signed Sept. 29, 2025, put transparency duties on large frontier developers, the law New York’s sponsors say their rewrite now more closely matches.

Put the dates side by side and the real deadline is obvious. Jan. 1, 2027 lands on California, New York and Colorado at once. A company that hires in all three, ships a chatbot and posts images on a big platform has several compliance projects with the same due date.

Sources

• Office of Governor Newsom: California's nation-leading AI framework just got stronger
• California Constitution Art. IV, Sec. 8
• LegiScan: CA SB 947
• CalMatters: On AI, Newsom gives labor only some of what it demanded
• LegiScan: CA AB 1883
• LegiScan: CA SB 951
• NY Senate: S8828
• Colorado Attorney General: AI