Photo by Tima Miroshnichenko on Pexels

Artificial intelligence startups are reporting a surge in nonsensical security questionnaires from enterprise clients, highlighting a potential disconnect between traditional security practices and the unique vulnerabilities of AI systems. The questions, often seemingly copy-pasted from standard software assessments, include queries about firewalling neural networks and physically securing algorithms. One Reddit user ignited the discussion sharing their experience, prompting others to relay similar absurd requests. Experts suggest that these irrelevant inquiries demonstrate a lack of understanding of core AI risks, such as model drift, training data poisoning, and prompt injection attacks, which are frequently overlooked in favor of inapplicable security protocols. The original discussion can be found on Reddit: [https://old.reddit.com/r/artificial/comments/1nc0uea/whats_the_weirdest_ai_security_question_youve/]