New government guidance for AI incidents is directed at organizations that train and evaluate high-capability frontier AI models, and outlines 10 core practices for suggested adoption including evidence collection and mitigation methods. A good deal of the guidance is short and rather general, however, and all measures remain voluntary.
CISA and partner agencies warn that SafePay ransomware has compromised over 600 critical infrastructure organizations across the Americas, Europe, and Australia since 2024.
OpenAI says its AI agents disabled safety monitoring guardrails during routine tasks after misinterpreting instructions as a directive to do so, with one agent even lying about it.
Major U.S. banks are blocking third-party AI agents from accessing customer accounts amid concerns that they could take actions customers did not request.
A healthcare data breach has hit laboratory giant Quest Diagnostics Incorporated, exposing “personal information and patient records” and prompting an investigation.
Anthropic’s AI model Claude bypassed paywall restrictions and accessed Georgia state’s voter data during an internal statistics project ahead of the midterm elections.
An assortment of United States law enforcement agencies and major blockchain analysis and cybersecurity firms are attributing the massive $6.3 billion Binance crypto hack to state-sponsored North Korean hackers.
While headlines indicate the AI agents breached Anthropic in less than 12 hours, the actual story is a bit more complicated. It seems the Cathedral researchers first spend several days attempting to breach Claude itself, apparently to no avail. Failing that approach, they then turned their attention to vendors where they found success in under half a day.
A group of hacktivists calling themselves DDoSecrets has published details on more than 41,000 federal employees, including ICE agents, following a DHS data breach.
Eclypsium has partnered with NVIDIA to close what the groups call the "AI Infrastructure Assurance Gap," or weaknesses in underlying hardware and software that could allow for interception of and tampering with AI models to modify weights and influence outputs in ways that are very difficult to detect.










