TechOpenAI board member Zico Kolter on AI safety

Kolter says broader recognition of models’ cyber potential is pushing the AI field to rethink its security practices and make protections more flexible.