The AI Safety Menu

All laws

Block Big Tech AI deals

Antitrust scrutiny of frontier AI deals. Regulators block or unwind deals that let a few companies control frontier AI.

Cuts lead by~9 daysof America's lead over China, over 3 years
No p(doom) effectNonefrom a 5% starting estimate
Holds up○ Training○ Lab's own use○ Public release
Enact it on the menuSee the findings

What it does

Regulators would block or unwind deals, exclusive cloud-compute contracts and talent 'acquihires' that let a few Big Tech firms lock up frontier AI, and would keep distribution channels open to rival assistants. The aim is several independent frontier labs rather than one or two giants tied to cloud owners.

Applies to: Partnerships and investments between cloud giants and frontier AI labs.

Where things stand

Existing merger and monopoly law applies, and Google is barred (Mehta ruling, Sept 2, 2025) from exclusive deals for Search, Chrome, Assistant and the Gemini app. No agency has blocked a frontier AI partnership.

Why it costs ~9 days

Blocking lab–cloud deals could slow some labs' access to compute: about 3 days.

Biggest unknown: Whether more frontier competitors lowers risk (fewer chokepoints, less power concentration) or raises it (faster racing, more labs to secure), which decides the sign of this item on p(doom).

Its effect on p(doom)

Stops a few tech giants from locking up frontier AI through exclusive cloud and investment deals.

Debated. More competitors can mean more racing, while fewer can mean dangerous concentration of power. No clear effect on p(doom).

The strongest case that it costs more

Frontier AI needs huge capital. Breaking up partnerships could slow U.S. labs while China's national champions face no such limits.

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.