Get the latest tech news
OpenAI rolls back ChatGPT’s sycophancy and explains what went wrong
Many organizations may also begin shifting toward open-source alternatives that they can host and tune themselves.
However, the update had an unintended side effect: ChatGPT began offering uncritical praise for virtually any user idea, no matter how impractical, inappropriate, or even harmful. By owning the model weights and the reinforcement learning process, companies can retain full control over how their AI systems behave—eliminating the risk of a vendor-pushed update turning a critical tool into a digital yes-man overnight. CEO Sam Altman has also previously stated the company plans to — in the coming weeks and months — release a state-of-the-art open source large language model (LLM) to compete with the likes of Meta’s Llama series, Mistral, Cohere, DeepSeek and Alibaba’s Qwen team.
Or read this on Venture Beat
