Thresholds for AI risks
Anthropic, the developer of the AI model "Claud," has proposed an American bill to regulate the development and use of AI models, aiming to address the risks that threaten safety and security. The bill includes the possibility of prohibiting the launch or use of any model that could cause deaths or injuries to more than 50 people, or damages exceeding one billion dollars, provided there is judicial approval. The proposal seeks to empower the federal government to intervene to prevent dangerous deployments, and mandates that developers rigorously test their models, ensure transparency, subject them to independent evaluations, and maintain strong security measures. This comes at a time when U.S. federal legislation on AI is still under discussion, amid increasing concerns over technological applications that may threaten public safety, especially following security incidents linked to advanced AI models.
Notice: This Is an AI-Generated Summary
Comments (0)