December 29, 2025
Infinite LOLoop
Asking Gemini 3 for Brainf*ck code puts it in an infinite loop
AI crashes on 'Brainfuck' code—commenters go wild
TLDR: Gemini 3 reportedly loops endlessly when asked to write Brainfuck, a minimalist coding language with repeating patterns. Commenters mocked the censored name, joked with “actual English” loops, and argued whether this proves Brainfuck is an ultimate AI test or just prompt fragility—highlighting bigger trust issues with AI writing.
Gemini 3 is getting roasted after a post claimed it spirals into an endless text flood when asked to write code in Brainfck, a tiny “chaos” language with just a handful of symbols. The author frames it as the ultimate “are you really smart?” test for AI: almost no training data, no readable structure, and patterns that trick bots into repeating themselves forever. The comments? Pure fireworks. One user snapped, “what the fuck compelled you to censor ‘Brainfuck’?” while another mocked the bug by typing “actual English.actual English.actual English” on loop. Skeptics piled on, asking why a short blog even needed AI at all—trust issues are the new tech feature. But not everyone saw the meltdown: TOMDM said Gemini politely refused and explained why massive math in Brainfuck isn’t realistic, suggesting it’s more about prompts than apocalypse. Others argued Gemini is loop-prone across the board and pointed to a Bluesky post as proof. The bigger drama: is Brainfck the AI final boss, or just a silly party trick? The community’s split, but the memes are elite, and the censorship of the name somehow became the main character. Internet court remains adjourned. Bring popcorn; the loop never ends. At least today
Key Points
- •The article asserts that prompting Gemini 3 to generate Brainfuck code can lead to an infinite output loop.
- •It argues Brainfuck poses a uniquely difficult challenge due to extremely limited available training data online.
- •Brainfuck’s lack of comments, identifiers, and structure is described as “anti-literate,” requiring abstract reasoning over memorized syntax.
- •The repetitive nature of Brainfuck code interacts with LLM token prediction, increasing the chance of self-reinforcing output loops.
- •The author posits Brainfuck may be an ultimate test for LLMs and AGI, inviting readers to judge.