The early bird gets the exploit.

CyberWire Daily covers fast-moving exploit activity, critical GitLab patching, CrowdStrike’s discovery of AI-driven attackers, and Interpol’s Africa-wide cybercrime takedown. Guest Danny Manimbo of Shellman argues CIOs shouldn’t wait for AI regulation clarity; they need governance, impact assessments, and human oversight now.

Key points

• Amazon saw PHP exploit attempts four days before disclosure, reinforcing that patch windows can shrink to hours.
• GitLab issued emergency patches for a maximum-severity RCE; Shadow Server tracks more than 2,700 exposed community instances.
• CrowdStrike found four undocumented threat actors using autonomous AI agents, with some attacks reaching objectives within an hour.
• CyberCartel targeted 400-plus Latin American financial organizations using multi-model AI and exposed nearly 4,500 operational records.
• Interpol’s Operation Serengeti 3 led to 1,400-plus arrests, 11,000 networks disrupted, and over $385 million in estimated losses.
• Danny Manimbo says effective AI governance needs ownership, inventories, impact assessments, and human override—not just compliance checklists.