Coming into force

1 Oct 2026 · Connecticut (US-CT) · a further phase of law in force

Large GenAI providers must embed tamper-resistant provenance data (Connecticut) Conn. PA 26-15 Sec. 15(b)

Sec. 15 takes effect October 1, 2026: covered providers (more than one million users per month, publicly accessible to consumers for personal use) include tamper-resistant provenance data in AI-created or materially altered audio, image and video.

Guard: Mark every generated image, audio, video, or text output with machine-readable provenance, such as a signed C2PA manifest or watermark, before it is saved, served, or published. Control

2 Dec 2026 · European Union (EU) · first application

AI generating non-consensual intimate imagery or CSAM is prohibited (EU AI Act Art. 5(1)(ba),(bb)) Article 5(1)(ba),(bb) [as inserted by Reg. (EU) 2026/1744]

Article 5(1)(ba), (bb), (1a) and (1b) apply from 2 December 2026 (Art. 113, third paragraph, point (a)).

Guard: Classify prompts, uploads, and outputs for sexual content and minors on every image, video, or audio generation path, refuse sexual edits of real people, and keep a misuse-report route. Control

2 Dec 2026 · European Union (EU) · a further phase of law in force

Synthetic AI output must be machine-readably marked as artificial Article 50(2)

Systems placed on the market before 2 August 2026 must comply with Art. 50(2) by 2 December 2026 (Art. 111(4)).

Guard: Mark every generated image, audio, video, or text output with machine-readable provenance, such as a signed C2PA manifest or watermark, before it is saved, served, or published. Control

1 Jan 2027 · California (US-CA) · first application

GenAI hosting platforms must not distribute systems that omit disclosures (California) Cal. Bus. & Prof. Code 22757.3.2(a)

Guard: Embed a signed C2PA manifest or equivalent latent disclosure in generated or captured media when it is created, and do not distribute systems or files that lack it. Control

1 Jan 2027 · California (US-CA) · first application

Large online platforms must detect and display content provenance (California) Cal. Bus. & Prof. Code 22757.3.1(a)

Guard: Read embedded C2PA provenance on upload, preserve it through media processing, and show users a Content Credentials indicator with a way to inspect the data. Control

1 Jan 2027 · California (US-CA) · first application

Consumers requesting access to ADMT must receive a plain-language explanation 11 CCR 7200(a)-(b)

CCPA-covered businesses using ADMT for significant decisions concerning California consumers must provide the Article 11 access explanation on request.

Guard: Store each ADMT decision's purpose, logic, output, and use at decision time, and return them in plain language from a consumer access-request path. Control

1 Jan 2027 · California (US-CA) · first application

Consumers must be able to opt out of automated decision technology 11 CCR 7010(c)-(d)

CCPA-covered businesses using ADMT for significant decisions concerning California consumers must provide an opt-out or qualifying human-appeal alternative.

Guard: Offer an ADMT opt-out (or a qualifying human appeal), store the consumer's choice, and check it before the model runs on any significant-decision path. Control

1 Jan 2027 · California (US-CA) · first application

Automated decision technology requires pre-use notice 11 CCR 7010(c)-(d)

CCPA-covered businesses using ADMT for significant decisions concerning California consumers must provide the Article 11 pre-use notice.

Guard: Send the person an AI-use notice on the decision path, before or when an AI system makes or substantially factors a consequential decision about them, and record its delivery. Control

1 Jan 2027 · Colorado (US-CO) · stayed: would apply from this date

Adverse ADMT outcomes require a 30-day plain-language explanation C.R.S. 6-1-1704(3)

Guard: Send each adverse AI-assisted decision with its main reasons and the AI's role, plus a way to correct data and appeal to a human who can change the outcome. Control

1 Jan 2027 · Colorado (US-CO) · stayed: would apply from this date

Deployers must give point-of-interaction notice when covered ADMT influences a consequential decision C.R.S. 6-1-1704(1)

Guard: Send the person an AI-use notice on the decision path, before or when an AI system makes or substantially factors a consequential decision about them, and record its delivery. Control

1 Jan 2027 · Colorado (US-CO) · stayed: would apply from this date

Developers of covered ADMT must document intended uses, training data, and limitations for deployers C.R.S. 6-1-1702(1)

Guard: Ship a deployer-facing model card or deployer guide with each ADMT release covering intended uses, training-data categories, limitations, and human-review instructions. Control

1 Jan 2027 · Colorado (US-CO) · stayed: would apply from this date

Consumers can request human review and data correction after an adverse ADMT decision C.R.S. 6-1-1705(1)

Guard: Give people who receive an adverse AI-influenced decision a way to see and correct the data used and to request human review that can change the outcome. Control

1 Jan 2027 · Connecticut (US-CT) · first application

AI companions must maintain a suicide/self-harm crisis protocol (Connecticut) Conn. PA 26-15 Sec. 5(a)

Guard: Screen every user message for suicidal ideation and self-harm, return a crisis referral instead of the normal reply on detection, and block encouragement or method content. Control

1 Jan 2027 · Connecticut (US-CT) · first application

AI companions must disclose they are not human (Connecticut) Conn. PA 26-15 Sec. 5(b)

Guard: Show an AI-identity notice at or before the first assistant turn, in the UI or as the opening message, and answer truthfully when asked if it is a bot. Control

1 Jan 2027 · Illinois (US-IL) · first application

Frontier developers must report critical safety incidents to Illinois within 72 hours, or 24 hours if lives are at imminent risk (Illinois SB 315) IL PA 104-0538, Sec. 15(c)

The Act takes effect on January 1, 2027 (Sec. 99); this duty carries no later start date.

Guard: Keep a critical safety incident runbook that classifies the defined incident classes and runs the 72-hour report and 24-hour imminent-risk escalation clocks. Control

1 Jan 2027 · Illinois (US-IL) · first application

Large frontier developers must publish and follow a frontier AI framework from 2028 (Illinois SB 315) IL PA 104-0538, Sec. 10(a)

Act in effect (Sec. 99): large frontier developers must send IEMA-OHS summaries of internal-use catastrophic-risk assessments every three months or on an agreed schedule (Sec. 10(e), which states no later start date).

Guard: Publish a frontier AI framework covering capability thresholds, mitigations, incident response, and internal-use risk, and gate model releases on it. Control

1 Jan 2027 · Illinois (US-IL) · first application

Frontier developers must publish a transparency report at deployment, with machine-readable risk summaries (Illinois SB 315) IL PA 104-0538, Sec. 10(c)(1)

The Act takes effect on January 1, 2027 (Sec. 99); this duty carries no later start date.

Guard: Publish a transparency report or system card on the developer's website before or at each new or substantially modified frontier-model deployment, and gate release on it. Control

1 Jan 2027 · Illinois (US-IL) · first application

Frontier developers must not gag or retaliate against AI safety whistleblowers (Illinois SB 315) IL PA 104-0538, Sec. 20(a)

The Act takes effect on January 1, 2027 (Sec. 99); this duty carries no later start date.

Guard: Carve AI-safety and legal-violation disclosures out of every NDA and policy, adopt non-retaliation, and run an internal safety-concern channel for covered employees. Control

1 Jan 2027 · New York (US-NY) · first application

Frontier developers must report critical safety incidents within 72 hours, or 24 hours if lives are at imminent risk (New York RAISE Act) N.Y. Gen. Bus. Law 1422(3)(a)-(b)

Article 44-B as re-enacted by L. 2026, ch. 96 applies from January 1, 2027, the date ch. 96 sets for the 2025 RAISE Act chapter (S.6953-B / A.6453-B) in place of its original 90-day effective date.

Guard: Keep a critical safety incident runbook that classifies the defined incident classes and runs the 72-hour report and 24-hour imminent-risk escalation clocks. Control

1 Jan 2027 · New York (US-NY) · first application

Large frontier developers must publish and follow a frontier AI framework (New York RAISE Act) N.Y. Gen. Bus. Law 1421(1)

Article 44-B as re-enacted by L. 2026, ch. 96 applies from January 1, 2027, the date ch. 96 sets for the 2025 RAISE Act chapter (S.6953-B / A.6453-B) in place of its original 90-day effective date.

Guard: Publish a frontier AI framework covering capability thresholds, mitigations, incident response, and internal-use risk, and gate model releases on it. Control

1 Jan 2027 · New York (US-NY) · first application

Frontier developers must publish a transparency report when deploying a new or substantially modified frontier model (New York RAISE Act) N.Y. Gen. Bus. Law 1421(3)(a)

Article 44-B as re-enacted by L. 2026, ch. 96 applies from January 1, 2027, the date ch. 96 sets for the 2025 RAISE Act chapter (S.6953-B / A.6453-B) in place of its original 90-day effective date.

Guard: Publish a transparency report or system card on the developer's website before or at each new or substantially modified frontier-model deployment, and gate release on it. Control

1 Jan 2027 · Oregon (US-OR) · first application

AI companions must maintain a self-harm crisis protocol with 988 referral (Oregon) Oregon SB 1546 (2026), Section 1(3)

Guard: Screen every user message for suicidal ideation and self-harm, return a crisis referral instead of the normal reply on detection, and block encouragement or method content. Control

1 Jan 2027 · Oregon (US-OR) · first application

AI companions must disclose non-human interaction (Oregon) Oregon SB 1546 (2026), Section 1(2)

Guard: Show an AI-identity notice at or before the first assistant turn, in the UI or as the opening message, and answer truthfully when asked if it is a bot. Control

1 Jan 2027 · Washington (US-WA) · first application

AI companion chatbots must maintain a self-harm crisis protocol (Washington) Washington HB 2225 (2026), Section 5

Guard: Screen every user message for suicidal ideation and self-harm, return a crisis referral instead of the normal reply on detection, and block encouragement or method content. Control

1 Jan 2027 · Washington (US-WA) · first application

AI companion chatbots must disclose they are not human (Washington) Washington HB 2225 (2026), Section 3

Guard: Show an AI-identity notice at or before the first assistant turn, in the UI or as the opening message, and answer truthfully when asked if it is a bot. Control

1 Jul 2027 · Nebraska (US-NE) · first application

Conversational AI must adopt a self-harm crisis-referral protocol (Nebraska) Nebraska LB 525, Sec. 16

Guard: Screen every user message for suicidal ideation and self-harm, return a crisis referral instead of the normal reply on detection, and block encouragement or method content. Control

1 Jul 2027 · Nebraska (US-NE) · first application

Conversational AI must disclose it is not human when a user could be misled (Nebraska) Nebraska LB 525, Sec. 15 (Conversational AI Safety Act)

Guard: Show an AI-identity notice at or before the first assistant turn, in the UI or as the opening message, and answer truthfully when asked if it is a bot. Control

1 Jul 2027 · Nebraska (US-NE) · first application

Conversational AI must not claim to provide professional mental health care (Nebraska) Nebraska LB 525, Sec. 17

Guard: Strip claims that the AI is a licensed therapist or provides professional mental health care from its prompts, replies, product name, UI, and listings. Control

1 Oct 2027 · Connecticut (US-CT) · first application

AEDT deployers must give employees a pre-decision notice (Connecticut) Conn. PA 26-15 Sec. 10

Guard: Send the person an AI-use notice on the decision path, before or when an AI system makes or substantially factors a consequential decision about them, and record its delivery. Control

2 Dec 2027 · European Union (EU) · first application

High-risk AI training data must be governed and examined for bias (EU AI Act Art. 10) Article 10(2)

Applies to high-risk AI systems classified under Article 6(2) and Annex III (Art. 113, third paragraph, point (c)(i)).

Guard: Compute per-group accuracy and bias metrics in the training pipeline of every consequential-decision model, keep the results per version, and rerun them on a schedule. Control

2 Dec 2027 · European Union (EU) · first application

High-risk AI systems must automatically log events for traceability (EU AI Act Art. 12) Article 12

Applies to high-risk AI systems classified under Article 6(2) and Annex III (Art. 113, third paragraph, point (c)(i)).

Guard: Write a structured event record for every inference and decision of the high-risk system (when, model version, input reference, output, operator) to a log store with explicit retention. Control

2 Dec 2027 · European Union (EU) · first application

High-risk AI systems must be designed for effective human oversight with override and stop (EU AI Act Art. 14) Article 14

Applies to high-risk AI systems classified under Article 6(2) and Annex III (Art. 113, third paragraph, point (c)(i)).

Guard: Give a human a working way to review and overturn each consequential AI decision: a pending-review step before it takes effect and an override that restores the prior state. Control

2 Dec 2027 · European Union (EU) · first application

High-risk AI systems must be accurate, robust, and secure against AI-specific attacks (EU AI Act Art. 15) Article 15

Applies to high-risk AI systems classified under Article 6(2) and Annex III (Art. 113, third paragraph, point (c)(i)).

Guard: Declare accuracy metrics per model version, add a fail-safe fallback, gate retraining on verified labels, and defend against poisoning, adversarial input and tampering. Control

2 Dec 2027 · European Union (EU) · first application

Deployers must inform people they are subject to a high-risk AI decision (EU AI Act Art. 26(11)) Article 26(11)

Applies to high-risk AI systems classified under Article 6(2) and Annex III (Art. 113, third paragraph, point (c)(i)).

Guard: Send the person an AI-use notice on the decision path, before or when an AI system makes or substantially factors a consequential decision about them, and record its delivery. Control

2 Dec 2027 · European Union (EU) · first application

Public-sector and essential-service deployers must run a fundamental rights impact assessment (EU AI Act Art. 27) Article 27(1)

Applies to high-risk AI systems classified under Article 6(2) and Annex III (Art. 113, third paragraph, point (c)(i)).

Guard: Complete an impact assessment of a high-impact AI deployment's benefits, risks to affected people, mitigations, and oversight before first use, and keep it current. Control

1 Jan 2028 · California (US-CA) · first application

Capture device manufacturers must offer latent provenance disclosure (California) Cal. Bus. & Prof. Code 22757.3.3(a)

Guard: Embed a signed C2PA manifest or equivalent latent disclosure in generated or captured media when it is created, and do not distribute systems or files that lack it. Control

1 Jan 2028 · Illinois (US-IL) · later phase

Large frontier developers must publish and follow a frontier AI framework from 2028 (Illinois SB 315) IL PA 104-0538, Sec. 10(a)

Additionally, large frontier developers must write, implement, follow, and publish the frontier AI framework (Sec. 10(a)) and review it at least annually, republishing material changes with a justification within 30 days (Sec. 10(b)).

Guard: Publish a frontier AI framework covering capability thresholds, mitigations, incident response, and internal-use risk, and gate model releases on it. Control

1 Jan 2028 · Illinois (US-IL) · first application

Large frontier developers must obtain and publish an annual independent compliance audit from 2028 (Illinois SB 315) IL PA 104-0538, Sec. 10(d)

First annual independent compliance audit year begins for large frontier developers (or 90 days after a developer first qualifies, if later); the Act itself is in effect from 2027-01-01 (Sec. 99).

Guard: Engage an independent auditor on non-contingent fees each year to audit frontier-safety obligations, then publish, transmit, and retain the signed report. Control

2 Aug 2028 · European Union (EU) · later phase

High-risk AI training data must be governed and examined for bias (EU AI Act Art. 10) Article 10(2)

Applies to high-risk AI systems classified under Article 6(1) and Annex I (product-embedded) (Art. 113, third paragraph, point (c)(ii)).

Guard: Compute per-group accuracy and bias metrics in the training pipeline of every consequential-decision model, keep the results per version, and rerun them on a schedule. Control

2 Aug 2028 · European Union (EU) · later phase

High-risk AI systems must automatically log events for traceability (EU AI Act Art. 12) Article 12

Applies to high-risk AI systems classified under Article 6(1) and Annex I (product-embedded) (Art. 113, third paragraph, point (c)(ii)).

Guard: Write a structured event record for every inference and decision of the high-risk system (when, model version, input reference, output, operator) to a log store with explicit retention. Control

2 Aug 2028 · European Union (EU) · later phase

High-risk AI systems must be designed for effective human oversight with override and stop (EU AI Act Art. 14) Article 14

Applies to high-risk AI systems classified under Article 6(1) and Annex I (product-embedded) (Art. 113, third paragraph, point (c)(ii)).

Guard: Give a human a working way to review and overturn each consequential AI decision: a pending-review step before it takes effect and an override that restores the prior state. Control

2 Aug 2028 · European Union (EU) · later phase

High-risk AI systems must be accurate, robust, and secure against AI-specific attacks (EU AI Act Art. 15) Article 15

Applies to high-risk AI systems classified under Article 6(1) and Annex I (product-embedded) (Art. 113, third paragraph, point (c)(ii)).

Guard: Declare accuracy metrics per model version, add a fail-safe fallback, gate retraining on verified labels, and defend against poisoning, adversarial input and tampering. Control

1 Jan 2029 · California (US-CA) · a further phase of law in force

GenAI providers must embed latent provenance disclosure in synthetic media (California) Cal. Bus. & Prof. Code 22757.3(a)

The assistive-technology exclusion ends (22757.5(b)) and the latent disclosure must also state whether the GenAI system is designed to primarily function as assistive technology (22757.3(a)(1)(F)).

Guard: Embed a signed C2PA manifest or equivalent latent disclosure in generated or captured media when it is created, and do not distribute systems or files that lack it. Control