Anthropic is advocating for a new legislative approach to AI safety. The proposed policy establishes a specific threshold for intervention, suggesting that a model could be blocked if it is deemed capable of causing the deaths of 50 or more individuals.
Despite the focus on preventing catastrophic outcomes, the proposal includes significant legal protections for AI developers. Under the suggested rules, federal agencies would be prohibited from blocking the deployment of a model unless they first obtain authorization from a judge.
What it means
This framework creates a tension between public safety and regulatory power. While it identifies a clear danger zone regarding loss of life, it limits the government's agility to act independently of the court system when managing high-risk AI releases.
