Claude attacks were 'Rorschach test' for infosec community, scaring former NSA boss

A batch of infosec headlines highlights accelerating, high-impact threats and the growing role of AI in both offense and defense. Reports at RSAC and elsewhere show AI agents (Anthropic’s Claude abuses, Google’s Gemini agents) being used to automate reconnaissance and attacks, prompting vendor responses (Okta agent controls) and calls for faster patching and stronger MFA. Multiple high-risk incidents include ransomware exploitation of a Cisco Secure Firewall zero-day (CVE-2026-20131) prior to disclosure, active exploitation of a critical Microsoft SharePoint vulnerability, abuse of Microsoft 1