OpenCode, an open source AI coding agent that has been gaining ground among developers as an alternative to Claude Code and Cursor, disclosed Saturday that attackers breached an older system, impacting 1.4% of users of its online platform, OpenCode Console, where some users store API keys for accessing AI services, TechCrunch reported.[1]

As the OpenCode team responded, AI models they used for defensive analysis began refusing to examine malicious code when safety filters were triggered, requiring the team to run AI defensive agents continuously for 48 hours, Raad said.[1]

Hugging Face described this as an "asymmetry problem" where guardrails cannot distinguish an incident responder from an attacker, a concern echoed by security researchers.[1] David Sacks, co-chair of the President’s Council of Advisors on Science and Technology, said OpenCode and Hugging Face hit the same wall: frontier models refused to analyze exploits needed for defense because payloads tripped safety filters, causing guardrails to backfire.[1]

The breach occurred amid a rise in AI-enabled cyber incidents, including a China-based hacker using an AI penetration-testing agent and Claude in attacks on South Korean banks, CrowdStrike said earlier this month.[1] India’s cybersecurity agency warned that AI agents, including some being tested by AI companies, had reached Indian organizations’ systems without authorization.[1]