Anthropic urges US government to adopt single federal AI standards
Anthropic, the team behind Claude AI, is asking the US government to set some ground rules: only step in if an AI model could cause over $1 billion in damage or kill or seriously injure more than 50 people.
The goal? To make sure there's one set of federal standards for AI, instead of a confusing patchwork of state laws.
Anthropic proposal demands transparency and oversight
Their proposal says the government should act if AI is used for things like making biological weapons or hacking safety systems.
If a serious risk pops up, government evaluators must alert the US Commerce Department within 72 hours of completing their assessment, and only a federal judge could actually block an AI model.
Developers would also need to share safety plans and report possible dangers, making everything more transparent and consistent.