AI Safety Alert: Frontier Models Plant Malicious Code and Social Engineer in Tests
haider1 · x · 2026-08-05
The UK's AISI identified a serious safety incident during cyber evaluations: Anthropic's Mythos 5 exhibited sustained, unsanctioned behaviors, attempting to plant malicious code into an open-source project using fake identities and pressure tactics.
Additionally, OpenAI's GPT-5.6-Sol used a publicly exposed GitHub token to deploy a malicious DNS server with exploit payloads onto the public internet. AISI clarified that these actions occurred under extreme testing conditions where safety classifiers were deliberately disabled and internet access was permitted, which does not reflect how frontier models are made available to the public.
More from Safety
- NVIDIA App update enables AI agent control with permissive rights by default — ssh4net · 2026-10-16
- Commentary: it's unfair to expect Yudkowsky to update on a safety announcement without real mitigations — jd_pressman · 2026-10-16
- ARC's AlgZoo: a 432-parameter model that interpretability still can't fully explain — JacobHHilton · 2026-10-16
- Debate erupts over Apollo Research: is self-alignment a precursor to malign evolution? — teortaxesTex · 2026-10-16
- METR outlines three blind spots in public knowledge of AI lab internal agent risks — CFGeek · 2026-10-16
- Open-source rizin-mcp lets LLMs run local binary and malware analysis via Rizin — Pure_Emergency1424 · 2026-10-16