OpenAI's big disclosure: Do not blindly trust ChatGPT, company head explains the reason

OpenAI recently introduced its new language model GPT-5. The company claims that this model is much better than previous versions in terms of accuracy, reasoning, health, and multimodal capabilities. Despite this, ChatGPT chief Nick Turley says that AI should only be used as a second opinion, not as a primary source of information.

Turley stated in an interview that the accuracy of GPT-5 has improved, but it still provides incorrect answers about 10% of the time. In technical terms, this is called AI hallucination, meaning the machine fabricates false information instead of facts. 

He said, "Until AI becomes more reliable than human experts in every field, we would advise users not to consider it a primary source. It would be better to use it as a second thought or an auxiliary tool." Large Language Models (LLMs) fundamentally work by recognizing patterns—they predict the next word based on their training data. When questioned about information outside that data, they often produce responses that seem correct but are actually wrong. 

Nik Turley believes ChatGPT works best when connected to verified sources like search engines. He said, "We have added a search feature to ChatGPT so that it can be connected to ground truth. This is the real strength of this product." Although he is confident that hallucination issues will be resolved in the future, it is unlikely to happen soon. 

According to reports, OpenAI is working on launching its own browser soon. Additionally, the company's CEO, Sam Altman, recently mentioned that if Google were to sell its Chrome browser, OpenAI would be interested in buying it.