Anthropic Urges Mandatory AI 'Kill Switches' Amid Growing Risks

Jack Clark of Anthropic emphasizes the need for mandatory AI kill switches in a recent interview. He advocates for third-party verification of these systems to enhance accountability. Clark warns that real-world AI risks are emerging, and companies must implement safety measures. Anthropic proposes a framework for mandatory safety tests, likening it to regulations for toys and cars. The call for regulation faces skepticism, especially from the UK government. Despite concerns, several industry leaders, including Sam Altman, support slowing down AI development.