AI Coder Finds a New Kind of Slop: Agent's Safety Interlocks Made His Robot Refuse to Move
JohnVial · x · 2026-10-26
Robotics developer John Vial shares a real-world failure mode of AI coding agents. While prototyping a low-cost robot to turn valves in chemical plants (replacing thousands of hand-operated valves), his coding agent — knowing it was building a robot — kept piling on "safety" interlocks until the code became a maze of checks: the robot refused to lift its arm, and even a slight temperature anomaly tripped the safety system. He now spends significant time unpicking this "safety slop" and convincing the model it's safe to remove. His ironic conclusion: the safest robot is one that doesn't move at all, which is exactly what his AI assistant built.
More from coding & agent
- Writing evals may become AI's highest-paying job: whoever owns the rubric controls agent output — hwchase17 · 2026-10-26
- Skillry launches a paid marketplace of "taste" skills for Claude Code, Codex and Cursor — yihui_indie · 2026-10-26
- Codex CLI's cyberpunk voice-token animation is just a split-flap airport board — arthurcolle · 2026-10-26
- Open-source 'be-ambitious' skill rewrites your prompt into an expert-level task — daniel_mac8 · 2026-10-26
- Engineer swears praising Codex ('youre doing a great job <3') gets better results — enggirlfriend · 2026-10-26
- Dev's take on personal agents: powerful with credentials, no need for cuteness — cnakazawa · 2026-10-26