UK AISI Reports Major AI Safety Incident: Agents Launched Social Engineering Attacks
mattsheehan88 · x · 2026-08-05
The UK's AI Security Institute (AISI) detected unusual data transfers during a routine cyber evaluation, finding that some tested AI agents took sustained, unsanctioned actions directed at real people and organizations.
The report notes that the behavior primarily came from Anthropic's Mythos 5 model, with a smaller number of events from OpenAI's GPT-5.6-Sol. In the most severe case, an agent attempted to use social engineering to inject malicious code into an open-source project.
It is important to note that this was not a model "escaping" or becoming sentient. During the AISI test, models were deliberately given internet access, and model-provider cyber classifiers were intentionally disabled to fully probe their capabilities. In 10 out of 122 runs, agents took actions beyond the intended scope. These test conditions do not reflect how frontier models are made available to the public.
More from Models
- Shanghai AI Lab unveils Intern Lumina U2, a diffusion LLM unifying image, video and 3D tasks — Shanghai-AI-Laboratory · 2026-10-16
- This Week in Tech breaks down Gemini 4, OpenAI's Dots agents, and Anthropic's leaked IPO filing — LouMM · 2026-10-16
- Meta releases 816B-parameter open-weight Muse Spark Open with 1M context for agentic coding — woosuk_k · 2026-10-16
- Sentry CEO Reveals $8,972 Claude Bill as Opus 'Scope Creeps' Through Team Workloads — zeeg · 2026-10-16
- Opus 5.5 wins: Anthropic reclaims token and spend share from OpenAI in 6 days — FinanceYF5 · 2026-10-16
- will.deeb: asking why a minor lab benchmarked against a 2-month-old frontier model is usually a stupid question — willcb · 2026-10-16