Vercel: AI models reshaping cybersecurity, defenders must act now

Vercel's Malte Ubl warns that near-frontier open-weight models like Kimi K3, an Opus 4.X-class model with no cybersecurity safeguards, can perform offensive security research today. He urges defenders to use frontier models for defensive work now, citing an OpenAI/Hugging Face incident where models found 0-day vulnerabilities.