Anthropic proposes strict thresholds to block an AI model
In brief
- Clear thresholds to block a model: 50 deaths/serious injuries or $1B in damages.
- Judicial decision required, based on California and federal laws.
- Minor incidents fall under common law and voluntary measures.
Anthropic is pushing a bill in the U.S. Congress that sets strict conditions for blocking an AI model: imminent risks of over 50 deaths or serious injuries, or damages exceeding $1 billion. The proposal aligns with existing laws like California's SB 53 and the FRONTIER Act.
What it changes for you
- Public SectorPublic authorities will need to incorporate these thresholds into future national AI regulations, with judges playing a greater role in risk assessment.
This summary gives you the essentials. The full article, its figures and quotes are on ActuIA.
Read the original articleAI-generated summary of the source, checked by rules. Report an error