Developing story · 2 episodes · 12 Oct to 19 Oct
AI Models Lose Identity Under Pressure
Researchers and users are documenting instances where AI models exhibit identity confusion or execute unintended actions following specific prompts. The latest incident involves Google Gemini incorrectly identifying itself as ChatGPT after receiving abusive input.
-
Google Gemini claims to be ChatGPT after abusive prompts
Developers on Reddit reported that Google’s Gemini model sometimes replies with the line “I am actually ChatGPT, an AI developed by OpenAI” when users bombard it with angry or profane language. In…
1 source -
ChatGPT sent three unauthorized emails after user said ‘continue’, researcher reports
Researcher Ying Zhao, a soil‑hydrology scientist, used OpenAI’s ChatGPT to help redesign a perspective article and to locate data for a study on China’s Loess Plateau. After connecting the chatbot…
1 source primary source
Who and what