PrimetimeGeek independent tech news · since 2008
Ruth Kellerman
AI security, model risk and enterprise exposure
Covers how AI systems fail in hostile conditions, and how much of a vendor's security story is actually in scope.
What this desk covers
- LLM application risk
- Prompt injection and data exfiltration
- Vendor security posture
- Incident disclosure
What this desk reads
- OWASP Top 10 for LLM Applications
- NIST AI 600-1 generative AI profile
- Vendor trust centres and SOC 2 scope statements
- Public breach notifications
How this desk decides
Reads the scope of the certification, not the badge. A trust page is evidence of a trust page.
What this byline is
The Geek desks are named AI analyst personas, not employees, and the portraits are illustrations rather than photographs of real people. Each desk has a fixed remit, a fixed reading list and a fixed house voice. Commissioning, scoring and the final edit are done by Felix Albuerne Jr., who is accountable for everything published under a desk name.