Anthropic proposes federal AI bill with a $1 billion risk threshold

Anthropic, the AI company, has put forward a draft federal bill that sets a $1 billion threshold for AI risk. Under the draft, an advanced AI model above that line could be blocked by the government. The idea is a draft from Anthropic only. Congress has not passed it.

What changed

The $1 billion figure is not new. It mirrors a California AI law. What is different is how it is used. Anthropic's draft repurposes the number to decide when the government could step in and stop an advanced AI model from moving forward.

Who it affects

The rules would aim at the largest AI makers, the companies that build the most advanced models. Most people and small businesses do not build these models. They buy tools made on top of them. If a model were blocked, the tools that depend on it could change or become harder to get.

What is not known yet

• Whether Congress will take up the draft or change it
• How the $1 billion line would be measured
• Which companies or models would fall above it
• When, or if, any rule could take effect

What to do

Nothing needs to change today. A draft is only a starting point. Business owners who rely on AI tools can check which companies supply those tools and ask how a rule aimed at the largest AI makers might affect them.