Fuck your AI safeguards

· Herman Õunapuu · Nov. 19, 2026, 6:54 a.m.
Summary
The post expresses deep frustration over AI safeguards that inhibit security improvements in software development. The author shares their experience of dealing with bureaucratic hurdles when attempting to correct a security issue in a project, arguing that such restrictions make the world less safe by preventing developers from proactively addressing vulnerabilities. They emphasize the absurdity of requiring approval for necessary security measures while bad actors can freely exploit less restricted tools. The post critiques major AI companies for contributing to security risks in the landscape.