Classic GPT-3 Fail: Mid-Sentence Aurora Explanation Erupts Into Profanity
flowersslop · x · 2026-08-13
A user shared a screenshot from the GPT-3 era: the model was seriously explaining the best time to see the aurora in Sweden, only to suddenly glitch mid-sentence (right at 'n-') and start violently repeating profanity (FUCK YOU). A classic, funny alignment failure case from early LLMs.
More from Fun
- Mathematicians Enter "Oh Shit" Phase: Rushing to Claim Credit and Monopolize AI Verification — RexDouglass · 2026-08-13
- AI Trader: Claude Makes 5 Stock Moves in a Day, Dumps ServiceNow — alejandroll10 · 2026-08-13
- ChatGPT Users Report Model Suddenly Cussing Like a Sailor — TheGame81677 · 2026-08-13
- Grok-4.6 Gets Starstruck by a Low GitHub ID — nbaschez · 2026-08-13
- Cognition Engineers Win DEF CON CTF with Perfect Blue Team — silasalberti · 2026-08-13
- DeepSeek-V4-Pro 'Excitedly Confused' by Its Own Abilities Sparks Memes — teortaxesTex · 2026-08-13