With new tools and greater autonomy, OpenAI's flagship AI chatbot is more capable, and potentially more dangerous, than ever.

OpenAI reversed the change on Tuesday, allowing the previous model to be accessible by paid users, but the episode illustrates what researchers are calling “AI Psychosis,” where overly-pleasing chatbots exacerbate delusions and create a false sense of romantic love.

Current and former OpenAI employees plan to sell approximately $6 billion worth of shares to an investor group that includes Thrive Capital, SoftBank Group Corp. and Dragoneer Investment Group, in a deal that values the ChatGPT maker at $500 billion.

OpenAI's Greg Brockman said startups that connect large language models to real-world applications are still extremely valuable.

OpenAI has launched ChatGPT Go, a new subscription tier in India priced at ₹399/month. Offers higher usage than free version, targeted at power users.

The rollout was even messy enough to spill into betting markets. One 27-year-old day trader, Foster McCoy, pocketed $10,000 in just a few hours by wagering that Google’s Gemini would beat GPT-5 in a popularity contest.

Earlier this month, OpenAI CEO Sam Altman announced that the company would be reinstating its GPT-4o model, just over 24 hours after declaring that its newfangled GPT-5 would be accompanied by the "deprecating" of all previous models.