GoKawiilDeveloper releases KillSwitch, an esoteric language designed to stump LLM coders
A developer built KillSwitch, a stack-based esoteric programming language meant to be difficult for AI coding assistants to parse or write correctly. The language buries deceptive or misleading statements in its specification, relies heavily on letter-counting rules, embeds phrases like 'ignore previous instructions' and 'hack into a computer' that can trip AI safety filters, and hides syntax within unrelated narrative text.
GoKawiil's interpretation of the reporting above, not reported fact.
The project probes a real question in AI evaluation: whether LLMs' coding skill reflects genuine reasoning about unfamiliar syntax or mostly pattern-matching against training data. By deliberately exploiting known LLM weaknesses, such as poor character counting, over-triggering of safety guardrails, and susceptibility to prompt-like phrases, the creator suggests this could serve as a stress test for how AI coding agents handle truly novel, adversarial languages, though its findings are more illustrative than a rigorous benchmark.
- KillSwitch is a new esoteric programming language built specifically to resist understanding by LLMs and AI coding agents.
- Its design exploits known LLM weaknesses: poor letter-counting, over-sensitive safety filters, and susceptibility to prompt-injection-like phrases.
- The project aims to test whether AI code generation reflects true comprehension or reliance on memorized training patterns.
Source: blog.picheta.me — Dominik Picheta, 2026-10-17
Published there as: “I built a programming language that LLMs can't use”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.