Leaked AI safety bill from Anthropic only lets government act if 50+ deaths or $1 billion are on the line

Anthropic, the creator of the Claude LLMs, has proposed a confidential draft of an AI safety bill to US lawmakers. According to Politico, the bill would allow government intervention only if an AI model poses a foreseeable, imminent risk of more than $1 billion in damages or the death or serious injury of more than 50 people. The potential risks include the creation or release of a biological weapon, the model evading its developer's controls, or a cyberattack.

The bill is based on Anthropic's Advanced AI Framework, a set of proposals that the company has been advocating for with federal and state lawmakers since June. Anthropic's spokesperson stated that the bill represents sensible policies they believe Congress should act on. However, critics argue that AI companies use warnings of theoretical, wide-reaching catastrophe to gain consent for their self-regulation and expansion, while doing little to address the existing harm AI causes in various industries and human psychology.

The proposed bill has been met with scrutiny, with Anthropic's safety lead recently quitting, stating that the world is in peril. Earlier this year, Anthropic's safety lead resigned, expressing concern over the potential dangers of AI. Just last week, OpenAI faced criticism from eight US state attorneys general for allegedly ignoring its own safety warnings for the sake of profit.

Critics argue that regulations like Anthropic's proposed bill give tech companies the illusion of compliance while actually restricting them on their own terms. For more information on AI, readers can refer to the author's running tally of the "dumbest things tech tycoons have said so far" and explore the best games featuring AI technology.