Comment by David Kertai

Policymakers should therefore be cautious about mandatory kill switches because they could create new security and reliability risks while offering only a limited solution to the broader challenge of controlling autonomous AI systems.