The approximately 150-day espionage campaign incrementally exfiltrated emails to cloud services.

Trump AI order proposes a 30-day voluntary review of frontier models before public release.

Project Glasswing partners discovered more than 10,000 vulnerabilities in its first month.

Study points out that AI has shattered the model of patching on a two- to four-week schedule.

The company was criticized after a blog posted that suggested law enforcement involvement.

While NIST upgraded the bug to 9.1, experts say teams must focus more on how attackers can exploit this flaw to gain VPN access.

Researchers say Russian-speaking group GreyVibe uses AI tools to scale cyberattacks on Ukraine.

The attack leverages a polyglot file, heavy obfuscation and fileless execution to evade detection.

IBM and Red Hat launch $5 billion effort to secure open-source software supply chains.

Single-turn attack success rates are not a reliable benchmark for model safety, Cisco concludes.