• Late Sunday, the House Energy and Commerce Committee released a bill that seeks to block states from passing regulation on AI “models, systems, or automated decision systems” for 10 years.

• First reported by 404 Media, the bill contains is a provision mandating that…

  …no State or political subdivision thereof may enforce any law or regulation regulating artificial intelligence models, artificial intelligence systems, or automated decision systems during the 10-year period beginning on the date of the enactment 11 of this Act.

• There are exceptions provided for laws and regulations that aim to “remove legal impediments” or “facilitate the deployment or operation” of AI systems, and those aimed at “[streamlining] licensing, permitting, routing, zoning, procurement, or reporting procedures.”

  • Essentially, all state laws on AI systems would be banned other than those which explicitly make it easier to develop and deploy AI.

• There has been a flurry of state-level legislative activity over the past year, with nearly 700 bills related to AI considered in 2024, according to Business Software Intelligence.

  • Of these, around 16% (113) were enacted into law, considerably higher than the 1-3% of bills that end up as enacted legislation in any given Congress.

    Subscribe now

How would it work?

• Automatic displacement under the Supremacy Clause. Because the bill expressly pre‑empts state law, any conflicting state AI rule becomes void and unenforceable the moment the federal law takes effect. States need not (and constitutionally cannot be required to) repeal their statutes; they simply cannot apply them.

• Enforcement would follow ordinary federal‑pre‑emption practice:

  • Defensive use: A company or individual targeted under a state AI rule could raise the moratorium as an absolute defense in state or federal court.

  • Offensive use: Regulated parties could seek declaratory or injunctive relief against state officials under Ex parte Young or 42 U.S.C. § 1983 to block threatened enforcement.

• The moratorium would be self‑executing—states could still leave their AI statutes on the books, but any attempt to apply them would quickly end up in federal court.

• Is it constitutional?

  • This likely turns on whether courts see the text as a legitimate “rule of decision” that preempts state regulation (likely permissible) or as an impermissible command telling states how to legislate or enforce (such as that doomed PASPA in Murphy vs. NCAA).

  • Because Congress offers no interim federal standards, challengers would have a credible Murphy‑style argument, making the provision at least litigation‑prone.

Commentary

• Responses to the proposed legislation have been mixed:

  • Rep. Jan Schakowsky (D-IL) told the Hill that it was a “giant gift to Big Tech.”

  • Sen. Ted Cruz (R-TX) told Business Insider that he was he was "very supportive of the principle," arguing that AI should be regulated via national standards.

  • Sen. Josh Hawley (R-MO) told Business Insider that he opposed the proposed moratorium, saying he didn't "want to tamp down on people's efforts to address" issues posed by AI.

Most notable, however, was not any new feature or capability. Instead, it was Anthropic’s own evaluation of the model, and upcoming ones, tucked away in the model’s system card. In the paper, the developers state that Claude 3.7 has shown increased capability to assist users in acquiring chemical, biological, and other dangerous materials.

How does Anthropic evaluate their models?

Anthropic’s model evaluation process is guided by its Responsible Scaling Policy (RSP). Before releasing a new model, the company conducts rigorous testing across key risk areas, including cybersecurity, autonomous capabilities, and potential misuse in high-risk domains like chemical and biological threats. Through automated assessments, red teaming, and human trials, testers attempt to probe the model’s limits and identify vulnerabilities.

With Claude 3.7 Sonnet, Anthropic has determined that existing safeguards (categorized as AI Safety Level 2 (ASL-2)) are still sufficient to prevent misuse. However, internal testing revealed that the model is approaching ASL-3 thresholds—the highest properly-defined safety level—which would trigger stricter security measures. In particular, evaluations of Claude 3.7 Sonnet’s ability to assist with high-risk areas, such as cybersecurity and biological threat modeling, “showed improved performance in all domains.”

“...based on what we observed in our recent CBRN testing, we believe there is a substantial probability that our next model may require ASL-3 safeguards”, the developers said, raising questions about capabilities and threats posed by upcoming model releases.

What are CBRN risks from AI?

Chemical, Biological, Radiological, and Nuclear (CBRN) risks are hazards that arise from the development and use of chemical substances, biological agents, radiological materials, or nuclear weapons. CBRN risks encompass a range of dangers, from chemical poisons and infectious pathogens to radioactive contamination and nuclear detonations.

CBRN risks from advanced AI manifest through the potential misuse of AI systems to develop or deploy these hazardous agents. AI's capabilities in data analysis, pattern recognition, and simulation can be exploited to design novel chemical compounds, engineer pathogens, or autonomously control CBRN weapons. In this way, AI could lower the barrier for malicious actors.

Although there are yet to be any instances where AI has directly contributed to the development or deployment of CBRN weapons, experiments have demonstrated that AI can inadvertently generate information regarding the creation of harmful materials.

In response to these emerging threats, there has been action on the part of policymakers and model developers to mitigate risks, including several pieces of legislation and private industry measures.