Archive · 12 Sep 2026
US Senators Weigh AI Law to Block Unsafe Models
US senators are discussing a new law about artificial intelligence.
The law could let the federal government stop some AI models from being released.
These models would have to be considered unsafe for that to happen.
Companies would be able to challenge the government’s decision in federal court.
Senators are still deciding how much power the government should have.
The proposal could also stop states from enforcing some of their own AI safety rules.
One concern is that criminals might use AI to help design nuclear or biological weapons.
The final shape of the law has not yet been decided.
Senate negotiators are considering giving the United States government power to block some AI models deemed unsafe.
Companies could challenge government blocking decisions in federal court.
Negotiators have not settled how much authority the government would receive.
The measure could prevent states from enforcing some AI-risk laws of their own.
Cited risks include bad actors using AI to design nuclear or biological weapons.
- Who
- United States senators, federal officials, companies developing AI models, and states.
- What
- Negotiators are considering a law that could let the federal government block certain AI models deemed unsafe.
- Where
- The proposal concerns the United States and would allow challenges in federal court.
- Why
- Supporters of the measure are addressing risks including the possible use of AI to design nuclear or biological weapons.
Supporters of Federal Oversight
Companies and State Authorities
Government power over unsafe models
Supporters of Federal Oversight
Federal officials could block the release of AI models judged unsafe, addressing risks such as assistance with designing nuclear or biological weapons.
Companies and State Authorities
Companies could challenge a blocking decision in federal court, while negotiators continue debating the limits of government authority.
National versus state regulation
Supporters of Federal Oversight
A federal measure could establish a national approach to certain AI risks.
Companies and State Authorities
The proposal could prevent states from enforcing some of their own laws governing those risks.
- Proposal
- A new AI law is under negotiation in the United States Senate.
- Government authority
- The federal government could gain power to block the release of certain unsafe AI models.
- Legal challenge
- Companies would be able to challenge blocking decisions in federal court.
- Unresolved issue
- Negotiators are still discussing the exact scope of government power.
- State laws
- Some parts of the measure could block states from enforcing their own laws on certain AI risks.
- Example risk
- A bad actor could potentially use AI to design nuclear or biological weapons.