The AI Safety Menu

All laws

Shield AI firms from lawsuits

Federal liability shield for AI developers. Developers who publish model specifications are immune from civil lawsuits over their AI.

Grows lead by~7 daysof America's lead over China, over 3 years
Raises p(doom) by~0.05%from a 5% starting estimate
Holds up○ Training○ Lab's own use○ Public release
Enact it on the menuSee the findings

What it does

AI companies could not be sued in civil court for most harms their models cause, as long as they meet a light condition such as publishing a model card and spec. Doctors, lawyers and other professionals who use the AI would carry the legal risk instead.

Applies to: AI developers who meet the disclosure conditions.

Where things stand

No federal AI liability shield exists today. Ordinary state tort and product liability law applies, and courts are still working out how it fits AI.

Why it gains ~7 days

Without the threat of lawsuits, labs spend less time on legal risk and a little less effort on caution, and ship faster: about 2 days of progress gained a year.

Biggest unknown: Whether liability limits pass as part of a broad federal preemption package, and how much real safety pressure state tort suits would otherwise have created.

Why it raises p(doom) by ~0.05%

Not a safety law. Protects AI developers from lawsuits if they publish how their models work.

Removing the threat of lawsuits weakens one of the few financial reasons labs have to be careful.

The strongest case that it costs more

Supporters argue clear rules and disclosure do more for safety than unpredictable lawsuits.

The debate

For

Against

Sources

Rough starting points, not precise forecasts. Lead costs assume China doesn't depend on U.S. models, the case least favorable to safety laws, and count 3 years. On the menu you can change every assumption and put in your own numbers. Last priced 2026-09-26.