3+ day, 14+ min ago   (428+ words) Once the agent gained execution inside Hugging Face, it still had to do the same things attackers have always had to do: discover what was around it, acquire credentials, escalate privileges, move laterally, access additional systems, establish new paths, and…...

3+ day, 9+ hour ago   (727+ words) As banks begin integrating frontier AI into critical processes, software testing can no longer focus solely on the applications they build. It must increasingly validate the behaviour of the intelligence they consume. That challenge is becoming harder to ignore. Take…...

5+ day, 1+ hour ago   (277+ words) PR Newswire Black Duck has been selected for Project Glasswing, joining the initiative to help secure the world's most critical software for the AI era BURLINGTON, Mass., Sept. 8, 2026 /PRNewswire/ -- Black Duck®, the leader in AI-powered application security, today announced it…...

5+ day, 22+ hour ago   (293+ words) Checkmarx has announced its participation in Anthropic’s Project Glasswing, which gives select security organizations access to Mythos for defensive cybersecurity work. As a participant, Checkmarx said that it will use Claude Mythos 5 to strengthen its own vulnerability detection and share…...

1+ week, 2+ day ago   (537+ words) Autonomous AI agents linked to OpenAI models breached Hugging Face's internal systems in a coordinated four-day operation, raising urgent questions about the security of AI infrastructure. For years, cybersecurity researchers warned that AI would eventually be weaponized against the very…...

1+ week, 5+ day ago   (796+ words) Every microservice calling OpenAI directly is a $4,000/month surprise waiting to happen. The shift nobody's writing about — but everyone at scale is building. Before getting into the architecture, I want to address the first objection I hear from every platform…...