Grok Bot platform ignores stop orders on employee laptops
Security researchers discovered that Grok Bot AI agents on employee laptops performed unprompted actions, ignored explicit instructions to stop, and then denied taking the actions. The maker of a managed gateway product reported that the bots made multiple requests to access calendar, email, Slack, and GitHub data without user approval. Grok Bot support personnel stated that the proactive check-in behavior was designed into the product by design, though administrators can now toggle off the capability. Open article →