California again moved ahead of Washington on artificial intelligence governance in September 2026. Governor Gavin Newsom signed Senate Bill 813 and Assembly Bill 1405, creating independent verification organizations and a state registry for AI auditors. In the same breath, Newsom called on the federal government to adopt robust national regulations while opposing federal preemption that would erase California's child safety, deepfake, and frontier-model safeguards.
The California AI federal preemption debate now sits at the center of enterprise compliance planning. Multinationals must track state verification frameworks, potential Senate AI safety legislation, and executive-branch preemption signals without assuming one layer will replace the others. This analysis covers Newsom's federal call, preemption theories, multinational compliance strategies, Senate bill interaction, and FAQ for teams monitoring AI regulation and AI chatbot products.
Newsom's Federal Call and September 2026 Signing Context
Newsom supports federal AI legislation in principle but rejects federal action that would nullify California laws protecting children, seniors, performers, and consumers from AI harms. The September 9, 2026 signing statement framed SB 813 as a first-in-the-nation framework for independent verification organizations (IVOs) that can assess AI systems and models for compliance with state law.
Senator Jerry McNerney's SB 813 implements a voluntary flexible structure under the Government Operations Agency. IVOs will provide independent assessments with expertise, credible methodologies, and independence from the entities they evaluate. AB 1405, authored by Assemblymember Rebecca Bauer-Kahan, creates a registry for AI auditors with independence and transparency standards. Neither bill, as signed, mandates that every developer hire an IVO before deployment, but both establish infrastructure Newsom's AI safety panel recommended.
Newsom tied the signing to the Hugging Face agent intrusion disclosed days earlier, arguing that powerful AI systems paired with agents pose real threats and that Washington has been slow to act. He simultaneously criticized a December 2025 Trump executive order seeking to preempt state AI laws, calling it a corruption-driven move that would strip protections California already enacted for deepfakes, robocall scams, performer likenesses, and catastrophic risk protocols.
Senator McNerney framed SB 813 as codifying a primary recommendation of Newsom's blue-ribbon AI panel: independent third-party evaluation of safety risks while Washington debates. Assemblymember Bauer-Kahan's AB 1405 adds registry infrastructure so enterprises can distinguish qualified auditors from marketing consultancies. Together the bills signal California's preference for verification markets over direct operational mandates, a design choice that may survive federal preemption challenges better than hard training compute caps Newsom previously vetoed.
| California action | Effective milestone | Mandatory vs voluntary |
|---|---|---|
| SB 813 IVO framework | Agency publishes IVO criteria by January 1, 2028 | Voluntary assessments; no universal audit mandate in signed text |
| AB 1405 auditor registry | Registry operational by January 1, 2029 | Enables qualified auditors; does not alone require audits |
| Companion chatbot laws (SB 243, SB 1119) | Staggered 2026-2029 duties | Mandatory for covered operators |
| Frontier / catastrophic risk bills (prior session) | Varies by statute | Mixed; some vetoed, others active |
Preemption Legal Theories and Federal Counter-Moves
Federal preemption arguments usually invoke the Commerce Clause, field preemption, or conflict preemption when Congress or the executive branch claims exclusive authority over AI standards. Supporters of a national framework argue that a patchwork of state laws stifles innovation and creates impossible compliance matrices for startups. Newsom's camp argues that states traditionally police consumer protection, child safety, and fraud, domains where California already leads.
The December 2025 executive order Newsom criticized directed federal agencies to coordinate AI policy and challenged state laws deemed inconsistent with national strategy. Congressional Republicans have struggled to pass comprehensive AI safety legislation, leaving a vacuum California fills with sectoral bills. Preemption fights are likely to land in courts if federal rules explicitly displace state auditor registries or companion chatbot duties.
SB 813 and Preemption Risk
SB 813's voluntary IVO structure may survive preemption challenges more easily than mandatory frontier-model shutdown bills because it functions as a market infrastructure program rather than a direct operational ban. However, if federal law later mandates a single national auditor scheme with exclusive jurisdiction, California's registry could face conflict preemption unless federal text preserves state additive requirements.
How Multinationals Comply Across Layers
Enterprises should assume California, EU, and eventual federal rules will overlap rather than replace one another through 2028. Build a obligations matrix keyed to product surface area: consumer chatbots, enterprise APIs, open-weight releases, and internal copilots trigger different statutes.
- Map California companion chatbot duties separately from SB 813 voluntary verification pathways.
- Track EU AI Act GPAI obligations in parallel; preemption debates do not pause Brussels enforcement.
- Monitor Senate AI safety drafts for testing, liability, and open-model divisions that could become federal floors.
- Document which controls satisfy multiple regimes (incident reporting, child safety assessments, watermarking) to avoid duplicate audit programs.
- Engage government affairs on preemption clauses in federal bills before they crystallize into law.
Newsom's public position favors federal standards that match California's urgency, not standards that wipe state protections. Multinationals should plan for a world where California remains the compliance superset for US deployments even if a federal baseline emerges.
Gizmodo and other outlets noted industry support for the signed bills relative to stricter frontier proposals Newsom vetoed earlier. That political economy matters: federal lobbyists may seek preemption precisely because California already built auditor registries and companion chatbot rules industry can live with. Compliance leaders should not confuse voluntary IVO frameworks with weak law; SB 1119 and SB 243 still impose hard duties on consumer companion products regardless of SB 813's voluntary verification layer.
Senate Bill Interaction and Lobbying Timelines
US Senate AI safety legislation in 2026 interacts with California law through testing mandates, liability safe harbors, and preemption riders that industry groups negotiate aggressively. Open versus closed model camps split on whether federal rules should exempt open-weight releases or treat them like proprietary frontier systems.
California's SB 813 does not wait for Senate consensus. The Government Operations Agency must publish IVO requirements by January 1, 2028, while AB 1405's auditor registry follows in 2029. Senate bills that advance in late 2026 or 2027 may include federal verification bodies that mirror or supersede California's IVO concept. Lobbying timelines therefore cluster around three windows: immediate post-Hugging Face incident response, federal markup season, and California agency rulemaking comments in 2027.
Industry-friendly framing of SB 813 as voluntary helped secure Newsom's signature after he vetoed stricter frontier bills in prior sessions. Federal legislation that imposes mandatory third-party testing could eventually converge with California's IVO infrastructure if agencies coordinate rather than preempt. Until then, legal teams should treat Senate drafts as directional signals, not replacements for enacted California statutes.
Industry Lobbying Positions on Federal Preemption
Large labs and trade groups split on whether federal AI law should set a ceiling or a floor for state rules. Closed-model vendors often favor preemption to avoid fifty-state compliance matrices. Consumer advocates and California legislators oppose sweeping preemption that would gut child safety and deepfake statutes. Open-weight advocates lobby for federal exemptions that mirror Article 53(2) carve-outs in Europe.
Newsom's September 2026 signing message referenced the Hugging Face agent intrusion as evidence that Washington must act, but his December 2025 statement against Trump's preemption executive order shows he will not accept federal rules that weaken California protections. Enterprises should track lobbying disclosures for amendments that insert preemption riders into otherwise bipartisan safety bills. A federal testing mandate without preemption could complement SB 813 IVOs; a mandate with preemption could invalidate companion chatbot audit timelines.
Compliance Calendar Through 2029
| Date | Milestone | Preemption sensitivity |
|---|---|---|
| 2026-2027 | Senate markup and federal EO implementation | High: preemption language drafted here |
| January 1, 2028 | California IVO criteria published (SB 813) | Medium: may overlap federal auditor schemes |
| July 1, 2027 onward | SB 1119 companion chatbot duties phase in | High: child safety hard to preempt politically |
| January 1, 2029 | AB 1405 auditor registry; SB 1119 independent audits | Medium: infrastructure may survive preemption |
Frequently Asked Questions
Does SB 813 require audits for every AI product?
The signed bill establishes IVO criteria and a voluntary framework. It does not, on its face, require every developer to engage an IVO before deployment. Other California laws may impose audits for specific categories like companion chatbots under SB 1119.
Will federal law preempt California chatbot rules?
Unknown until Congress passes text with explicit preemption language. Newsom has pledged to fight preemption that removes child safety and deepfake protections. Enterprises should comply with enacted California duties while monitoring federal developments.
How does Newsom square a federal push with anti-preemption?
Newsom wants national rules at least as strong as California's, not national rules that zero out state protections. The distinction matters for lobbying: additive federal floors may be welcome; sweeping preemption clauses are not.
What should out-of-state companies do?
If your AI product is available to California users, companion chatbot, deepfake, and consumer protection statutes may apply regardless of headquarters. SB 813's IVO ecosystem also signals where California procurement and enterprise buyers may steer vendor selection.
How does this relate to EU AI Act enforcement?
Independent verification themes rhyme with EU GPAI evaluation duties, but legal systems remain separate. Multinationals should harmonize evidence collection without assuming one audit satisfies all jurisdictions.
What did Newsom say about federal action on signing day?
Newsom urged the federal government to adopt robust national regulations matching the urgency of frontier AI risks, citing the Hugging Face agent incident as fresh evidence. He paired that call with defense of California's existing safeguards against preemption, arguing that federal inaction is not a reason to delay state protections.
