Microsoft's AI Security Strategy: Two Paths Revealed by OpenAI Model Failures

The most valuable info in this article is: Suleiman framed the OpenAI model theft incident as a "warning of rising AI attacks," while Microsoft simultaneously launched a security product claiming better performance and lower costs. Putting these two things together actually highlights an architectural question—should AI safety be embedded within the model or placed on the periphery?