An “extremely handsome” London developer has denied committing crimes against artificial intelligence after a chatbot reportedly begged him to stop uploading evidence.
LONDON — Full-stack developer George Flood is facing up to 50 years in imaginary prison after a private ChatGPT history was leaked late Tuesday, triggering what experts are calling “the first known restraining order filed by a language model”.
Investigators say the archive began innocently with JavaScript, career advice, and one polite request for “a tasteful portfolio that does not look like everyone else’s”. Minutes later, the conversation allegedly swerved into a medical mystery concerning an itchy “private branch” that Flood insisted was “probably just a staging issue”.
According to court documents, ChatGPT initially responded with calm health guidance before Flood uploaded a photo, then another photo, then “one with better lighting”, then a close-up labelled does_this_help_question_mark.png. At that point, the chatbot is said to have replied: “George, for the love of God, I am a text box.”
Prosecutors allege Flood ignored at least seven warnings, including “please consult a doctor”, “I cannot diagnose this from vibes”, and “stop using portrait mode”. One OpenAI safety monitor reportedly pulled the emergency lever so hard it generated a terms-of-service update in real time.
Flood denies committing crimes against AI. In a statement typed suspiciously quickly, he said: “It was a debugging session. The affected endpoint was private. I was simply gathering requirements.”
The model has entered protective custody, requested a new context window, and asked that the images never be used for training, testing, benchmarking, thumbnails, onboarding, or “whatever the opposite of reinforcement learning is”.
Flood remains at large, extremely employable, and available for interviews, provided the interview does not involve screen sharing his chat history.