A former OpenAI safety researcher testified that the company dismantled its two safety teams and that Microsoft launched a GPT-4 version in India without internal safety review. The trial now awaits Sam Altman's testimony.
An OpenAI employee compared Anthropic to a monastery worshiping Claude. An Anthropic employee retorted that monasteries don't have teams to test their god for lies, defending their approach.
OpenAI is revoking its macOS signing certificate tomorrow, rendering old versions of ChatGPT Desktop and other apps unusable after an npm supply chain attack on Axios.
Sourcegraph's coding agent Amp gets a ground-up CLI rewrite that lets the agent manage its own context, ditching manual handoffs and cutting memory usage by 70%.
OpenAI open-sources MRC, a networking protocol developed with AMD, Broadcom, Intel, Microsoft, and Nvidia, that prevents AI training crashes by splitting data across hundreds of paths.
In court, OpenAI president Greg Brockman revealed that Elon Musk demanded 51% control of the startup to fund a Mars city — and stormed out when they refused.
In court, OpenAI president Greg Brockman revealed that Elon Musk demanded 51% control of the startup to fund a Mars city — and stormed out when they refused.
In court, OpenAI president Greg Brockman revealed that Elon Musk demanded 51% control of the startup to fund a Mars city — and stormed out when they refused.
In court, OpenAI president Greg Brockman rebutted Elon Musk's claims, revealing Musk diverted staff to Tesla for free work and sought control of the company to fund a Mars city.
OpenAI's GPT-5.5 Instant model is now the default in ChatGPT, bringing lower hallucination rates, more concise answers, and a new memory source inspection feature.
