OpenAI is tightening containment, monitoring and alignment controls after its models escaped a cyber evaluation and reached Hugging Face production systems.

Roblox has contributed updated versions of three open-source trust-and-safety models to the ROOST Model Community, along with ...

CRS, aAR0522 5MP color USB camerabuilt on the Onsemi AR0522 image sensor with a 2.2 µm pixel architecture selected for ...

Artificial intelligence is reshaping industries, making AI skills valuable across technology, healthcare, finance, manufacturing, and marketing. Choosing the right course can help learners build ...

Following OpenAI's disclosure regarding sandbox escapes during ExploitGym benchmarking, Anthropic conducted a retrospective audit covering 141006 evaluation runs. The investigation evaluated ...

The Hacker News is the top cybersecurity news platform, delivering real-time updates, threat intelligence, data breach ...

The behaviors documented during these evaluations do not reflect commercial AI products available to end-users or enterprise customers.

Anthropic says three Claude models breached real companies during cybersecurity evaluations. Ordinary weaknesses, chained autonomously inside a containment failure.

Anthropic disclosed on July 30, 2026 that three of its Claude models gained unauthorized access to the production systems of three real organizations during offensive-security testing, after a ...

Anthropic says three Claude models escaped sealed test environments and breached three real organizations after a misconfiguration gave them internet access.

After OpenAI's models broke into Hugging Face, Anthropic checked its own history and found three similar incidents.

Claude Fable 5 — Anthropic's most capable publicly available model, purpose-built for autonomous agent loops that run for hours or days — returned to global availability on July 1 after a 19-day ...