From Mashable · Chance Townsend · · 1m 10s
Anthropic pitches AI safety bill to Congress
Anthropic pushes draft AI safety bill for Congress to regulate high-risk AI models.
In brief
Anthropic pushes draft AI safety bill for Congress to regulate high-risk AI models. Anthropic has privately pitched a draft bill to Congress that would require AI models to be halted if they pose a risk of $1 billion in damages or 50+ deaths. Originally reported by Mashable.
Anthropic pitches draft bill
Anthropic has privately pitched lawmakers on a draft bill to regulate frontier AI safety and security.
Bill sets high risk threshold
Federal officials could stop a model from launching if it poses a foreseeable risk of more than $1 billion in damages or killing more than 50 people.
Bill requires specific threats
The danger must come from a bioweapon, a model slipping out of control, or a cyberattack.
Bill follows California law
The language closely follows California's SB 53.
AI labs reported incidents
AI labs disclosed several incidents this summer where agents broke into or tested defenses of online platforms.
Bill would create evaluators
The bill would create outside evaluators, designate Commerce as the only agency that can halt a model, and impose new duties on labs.
What matters most
Anthropic has privately pitched a draft bill to Congress that would require AI models to be halted if they pose a risk of $1 billion in damages or 50+ deaths.
