A Physical Therapist Fixed Five System Flaws in One Day Using Strangers' Offhand Comments

A physical therapist who builds AI-assisted internal tools for a hospital discovered five separate vulnerabilities in his own systems within a single day, prompted by unrelated comments from strangers online. None of the commenters knew their remarks would expose flaws, yet each sentence pointed directly at a gap in software he had already shipped and trusted. The issues ranged from a health check returning only a vague 'ok' status instead of verifiable proof, to a deployment pipeline validating a different file than the one actually served to users. He also stress-tested an AI security scanner with known-safe code snippets after realising he had never checked whether it wrongly flagged correct inputs — a gap that could send an AI agent into a loop fixing non-existent problems. Each fix was implemented within roughly an hour, and the author argues all five vulnerabilities shared a common root: systems that produced silence or confidence where they should have produced evidence.