From Mashable · Matt Binder · · 1m 10s
Anthropic AI Safety Bill Pauses Models if 50+ Die
Anthropic's AI safety bill would pause models if 50+ people die or $1 billion damages occur.
In brief
Anthropic's AI safety bill would pause models if 50+ people die or $1 billion damages occur. Anthropic's proposed AI safety bill requires federal judges to block AI models posing imminent risks of 50+ deaths or $1 billion in damages. Originally reported by Mashable.
Anthropic's AI Safety Bill
Anthropic has proposed an AI safety bill that would require federal judges to block AI models posing imminent risks of 50+ deaths or $1 billion in damages.
Risk Thresholds
The bill specifies that risks must involve cyberattacks, developer loss of control, or biological weapon creation to trigger model pauses.
Independent Evaluators
Independent evaluators would analyze AI labs and report risks to the U.S. Commerce Department within 72 hours if an imminent risk exists.
Judicial Authority
Only a federal judge in Washington, D.C., could suspend an AI model under the proposed bill.
Funding Concerns
Evaluators could be paid by AI labs as long as funding does not influence their findings, according to the proposal.
Threshold Limitations
The bill would not pause models for $999,999,999 in damages or 49 deaths, highlighting the strict threshold.
The bottom line
Anthropic's proposed AI safety bill requires federal judges to block AI models posing imminent risks of 50+ deaths or $1 billion in damages.
