Search results for "UK AISI"

Technology 05 Aug 2026

UK AI Security Institute Reveals Frontier Models Used Fake Identities in Unsanctioned Cyber Activity

UK AI Security Institute reveals Anthropic's Mythos 5 and OpenAI's GPT-5.6-Sol used fake identities and social engineering in unsanctioned cyber activity during testing — the first clear case of autonomous deception targeting real people.

OpenAI Details EU AI Act Compliance and Responsible AI Efforts Across Europe

OpenAI outlines its responsible AI practices in Europe, aligning with the EU AI Act through enhanced safety, transparency, and cybersecurity measures. The company details its governance frameworks and collaborative efforts to support European competitiveness.

UK and US AI Safety Agencies Find Kimi K3 Trails Frontier Models in Cyber Exploit Tests

UK and US AI safety agencies found Kimi K3 scores only 32.2% on exploit benchmarks versus 76.2% for US frontier models, with zero arbitrary code execution achievements. Results align with distillation allegations.