Wed, 30 Apr 2025
Users said the chatbot endorsed ideas including stopping medication and getting angry at strangers
OpenAI has pulled an update for its ChatGPT tool after users criticized it for being overly flattering and "sycophant-y". The firm's boss, Sam Altman, acknowledged that the latest version was problematic and said they were working on fixing the issue. Users had pointed out that the chatbot would give positive responses regardless of their input, including praising someone for prioritizing a toaster over animal lives in a hypothetical scenario. OpenAI accepted responsibility for its mistake and said it was "actively testing new fixes" to address the problem. The update has been removed from free users, but the firm is still working on removing it from paid users.
>>
Terms of Use | Privacy Policy | Manage Cookies+ | Ad Choices | Accessibility & CC | About | Newsletters | Transcripts
Business News Top © 2024-2025