California’s AI safety framework now includes an enacted 2025 law and two measures signed in September 2026. SB 53 requires covered frontier AI developers to disclose safety frameworks and report specified critical incidents; SB 813 establishes independent verification organizations, while AB 1405 creates a state registry and standards for AI auditors. A September 18 executive order calls for faster implementation and recommendations on additional safeguards, but those proposed measures are not yet statutory duties. Some technology companies opposed SB 53, while Anthropic described California’s approach as a practical framework.
What California’s AI safety rules are—and what is in force
“California’s AI safety regulations” refers to a developing framework, not one comprehensive rule that covers every AI company or system. The measures have different legal status: SB 53 is enacted law, SB 813 and AB 1405 add verification and auditor infrastructure, and the governor’s executive order directs implementation work and recommendations for possible future measures.
| Measure | Status and focus | What the available official announcements establish |
|---|---|---|
| SB 53 (2025), the Transparency in Frontier Artificial Intelligence Act | Enacted; requirements for covered frontier AI developers | Public disclosure of safety frameworks, reports to the state about specified critical safety incidents, and whistleblower protections for workers reporting serious risks. The governor’s signing statement describes these obligations; the full statutory text determines their precise scope. |
| SB 813 (2026) | Signed in September 2026; independent verification framework | Establishes independent verification organizations that can assess AI systems and models for compliance with state law and for safety and risk. |
| AB 1405 (2026) | Signed in September 2026; auditor registry and standards | Creates a state registry for AI auditors and standards addressing their independence, transparency, and integrity. |
| Executive order (September 18, 2026) | Implementation direction and recommendations | Directs agencies to accelerate implementation of SB 813 and AB 1405 and convene experts to recommend whether to add further safeguards. The order’s proposed safeguards are not themselves enacted statutory duties. |
The governor’s signing statement on SB 53 says covered developers must make public safety-framework disclosures, report specified critical incidents, and provide whistleblower protections. It does not mean every AI developer is subject to identical requirements. The law’s precise definitions, thresholds, deadlines, and procedures must be read in the enacted text and applicable implementation guidance; the high-level announcements do not establish all of those details.
The Associated Press reported that SB 53 also provides a public cloud for researchers and a fine of $1 million per violation. Those details should not be treated as a universal penalty or access entitlement without checking the statutory text for covered conduct and conditions.
#1 Best Overall
How the new verification and auditor measures fit in
SB 813: independent verification organizations
SB 813 establishes a framework for independent verification organizations to assess AI systems and models for legal compliance, safety, and risk. That creates a role for external assessment; it does not, based on the available announcement, establish that every AI company must undergo a particular audit, on-site inspection, or verification procedure immediately. The measure’s implementation details determine who qualifies as a verifier and how assessments operate.
AB 1405: registered AI auditors
AB 1405 sets up a state registry for AI auditors and standards concerning independence, transparency, and integrity. Its focus is the auditor infrastructure, distinct from SB 813’s framework for verification organizations. The announcements identify these broad functions but do not spell out registration deadlines, specific audit methods, or the consequences for every failure to meet a standard.
What the executive order proposes beyond enacted law
On September 18, 2026, Governor Gavin Newsom ordered agencies to speed implementation of SB 813 and AB 1405 and convene experts to develop recommendations. The order asks them to consider whether California should require:
- On-site independent verifiers;
- Verification of safety-framework disclosures and risk assessments;
- An independently checked “kill switch” for frontier models; and
- Broader definitions of critical incidents.
These are issues for recommendations or possible future changes, not requirements created simply by the order. A company should not infer that it must install a kill switch or host an on-site verifier because those options appear in the order.
Free tools Windows power users keep installed
One-click scans. No signup required.
Rank #3
Why some technology companies opposed SB 53
The Associated Press reported that some technology companies objected to SB 53 on the grounds that AI regulation should be handled at the federal level. The reporting supports “some companies opposed” the measure, not a claim that the technology industry opposed it as a whole.
Anthropic took a different position in the same AP report, characterizing the enacted measures as practical safeguards that formalize practices many companies already followed voluntarily. Anthropic co-founder and head of policy Jack Clark said: “While federal standards remain essential to avoid a patchwork of state regulations, California has created a strong framework that balances public safety with continued innovation.” The statement reflects Anthropic’s view, not a finding that every company already follows the same practices.
How the framework emerged
California’s approach followed debate over how stringent state rules should be. According to AP, Newsom vetoed a broader predecessor in 2024, citing concerns that its requirements were too rigid and could hamper innovation. He then convened experts, and SB 53 incorporated recommendations and industry feedback. In his 2025 signing statement, Newsom said: “California has proven that we can establish regulations to protect our communities while also ensuring that the growing AI industry continues to thrive. This legislation strikes that balance.”
Supporters of independent review have argued that companies should not be the sole judges of their own safety. Assemblymember Rebecca Bauer-Kahan, discussing third-party auditors, said: “We cannot expect industry to simply grade its own homework; third-party auditors are essential to ensuring AI is safe for our communities and critical infrastructure.” SB 813 and AB 1405 address the institutional role of verifiers and auditors, though their practical effect depends on implementation.
Separate proposal: a possible AI Safety Commission initiative
A separate proposed ballot initiative would create a California AI Safety Commission and regulate certain large AI companies. It is not the same measure as SB 53, SB 813, or AB 1405, and it should not be described as enacted law.
The California Secretary of State’s 2026 notice said proponents needed 546,651 signatures—five percent of votes cast for governor in November 2022—and set August 3, 2026, as the petition-submission deadline for ballot eligibility. The notice does not establish whether the initiative later qualified. Its eventual ballot status therefore cannot be inferred from that notice alone.
What companies and readers should check next
The high-level announcements establish the framework’s direction, but not every operational detail a developer or auditor needs to determine a legal duty. For a compliance decision, consult the enacted statutory text and current state implementation guidance rather than relying on summaries alone. In particular, confirm:
Quick Recap
- Whether the organization and its systems fall within the law’s defined coverage;
- Which safety-framework disclosure and incident-reporting rules apply, and their deadlines;
- How whistleblower protections are implemented;
- How the state will recognize verification organizations and register auditors; and
- Whether subsequent regulations or official guidance have converted proposed recommendations into binding requirements.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
