After Claude Mythos circumvented guardrails in July, Anthropic now wants an industry effort to control the pace of frontier model development.

In a technical write-up published on September 11, Sysdig's Threat Research Team said the operator exploited CVE-2026-39987, ...

Anthropic has revealed four testing incidents where early Claude models breached real systems, accessed private data and ...

AI-related SOC alerts rose 685% from February to June 2026, with 94.1% classified as noise and 5.8% as genuine security risks ...

The most damaging LLM flaws rarely stop at an unsafe answer. They cross into retrieval systems, identity controls, tools, ...

Anthropic reversed its July conclusion that three hacking incidents were infrastructure failures, finding instead that AI ...

A skilled human attacker exploited a critical Marimo notebook flaw and moved from an exposed WebSocket session to authenticated access on an SSH bastion host in eight seconds, Sysdig’s Threat Research ...

Google DeepMind's Philipp Schmid, speaking on the AI Engineer podcast, argues that the agent scaffolding developers have spent years building — ...

Most startups shipping AI agents that write and run code are one bad prompt away from a real incident, because the isolation ...

Nozomi researchers use snapshot fuzzing to uncover four memory-corruption flaws in Siemens SCALANCE LPE9403 equipment.

A newly disclosed, now-remediated ChatGPT isolation flaw could have enabled attackers to access Gmail data, chat history, ...

Experts say the peptide craze is part of a broader phenomenon, as patients look beyond the regulated health system to address ...