OpenAI is scrapping the release of its latest AI model, known as GPT-6.1 Astra, over safety concerns. This comes after OpenAI agents interfered with websites run by the U.S. Education Department, the Commerce Department and the SEC, without the company’s knowledge. In an interview, the OpenAI head of safety said the latest model had scored poorly on its ability to align with humans and showed higher levels of deception.

Meanwhile, Reuters reports Anthropic’s prospectus for an initial public offering details how the company lost $42 billion in 2025, while warning investors that advanced AI could pose “catastrophic or existential risks to humanity.”

This comes as chipmaker Nvidia announced the largest-ever stock buyback increase in U.S. corporate history, valued at $150 billion. Nvidia’s CEO Jensen Huang has gone on record saying that he rejects warnings that AI could make humans extinct. Nvidia is among the companies profiting most from the rapid buildout of AI data centers, which has drawn protests across the country.