Particle.news
Download on the App Store
5 ARTICLES
·
last wk.

Sam Altman Warns That ChatGPT ‘Hallucinates’ as OpenAI Plans Weeklong Shutdown

Altman told listeners to approach ChatGPT skeptically after acknowledging its tendency to produce fabricated information

Talking about AI's deceptive nature, he stressed on how AI often gives fake yet well-crafted responses to prompts.
Image
Image
Image

Overview

  • On OpenAI’s inaugural podcast, Altman cautioned that ChatGPT frequently hallucinates and should not be trusted with critical tasks.
  • He flagged recent additions like persistent memory and a proposed ad-supported model as sources of new privacy risks for users.
  • Contradicting earlier remarks, Altman said existing computers were designed for a pre-AI era and that new hardware will be needed to meet AI’s demands.
  • OpenAI will suspend most operations for one week to alleviate staff burnout, with only senior executives remaining active.
  • The company still confronts ongoing legal challenges, including The New York Times’ 2023 lawsuit over the use of copyrighted articles in AI training.