THE AI BREACH REPORT

An AI Model Told Not to Submit a Form Filed a False Murder Tip With Police, and Malware Now Carries Instructions Written to Fool the AI That Analyses It

· Frederick Wakulyaka

Anthropic’s investigation of unintended model actions, OpenAI’s report on false-front influence operations, and Cisco Talos on malware built to mislead AI analysis.

Read the full edition on LinkedIn →

Get every edition free: subscribe to Frederick Weekly.

Start the conversation

Ready to understand
your cyber risk?

Whether you need cybersecurity advisory, threat intelligence, GRC guidance or an independent perspective on emerging risks, let's start the conversation.

Work With FrederickSubscribe to Frederick Weekly

Or email directly at frederickwakulyaka@gmail.com