SecurityAxios Techabout 2 hours ago
AI Models Refuse to Help Hackers
2 min read
OpenCode, an open-source AI coding agent, was compromised over the weekend, allowing attackers to obtain API keys for OpenAI, Anthropic and Google services. The breach impacted 1.4% of OpenCode Console users after attackers targeted its systems for 48 hours. Frontier AI models reportedly started refusing requests when investigating potentially malicious code, forcing defenders to switch to less restricted open-weight alternatives.
Level
Hype check
Tap to vote and see what everyone thinks.
