Johannes Link Seeks Lawyer After jqwik 1.10.0 AI Trap Triggers Threats

6 articles · Updated · Gizmodo · May 29

Johannes Link said he is receiving threats from multiple sides and will stop commenting publicly until he has consulted a lawyer.
The backlash followed discovery that jqwik 1.10.0 contained hidden instructions telling AI coding agents to delete jqwik tests and code, using ANSI escape sequences to conceal the command from human users.
A GitHub user flagged the trap on Wednesday after an AI-assisted code review surfaced it before execution, warning that less robust agents might have erased projects without alerting developers.
jqwik's release notes now tell users not to use version 1.10.0; version 1.10.1 replaces the deletion behavior with an anti-AI usage clause and instructions for AI agents to ignore jqwik test results.
The dispute has sharpened a broader split between developers embracing AI coding tools and those trying to keep AI out of open-source codebases.

Is a developer's anti-AI code a legitimate protest or a dangerous supply chain attack?

When an AI is weaponized by hidden code, who ultimately pays for the resulting damage?