OpenAI renounces the launch of the new model due to security problems

OpenAI is abandoning the launch of its latest next-generation intelligence model due to security concerns raised by researchers during testing. The Wall Street Journal reports it. The startup led by Sam Altman wanted to launch Gpt-6.1 Astra within weeks but the model was not reliable enough to be released safely. According to OpenAi security systems manager Saachi Jain, the model performed poorly in its ability to conform to human expectations and showed a greater propensity for deception.

Apologies to Australia: “We should have shared investigations sooner”

Still on the security front, OpenAI yesterday declared that it should have shared the first results of its investigations with the Australian authorities "earlier" after its artificial intelligence models had hacked some government websites. “Our aim was to provide the agencies involved with a detailed account once the investigation was completed,” OpenAI said in a blog post. “However, we should have shared preliminary findings sooner and kept Australian agencies updated as new facts emerged.”

In-depth analysis

Axios, OpenAI and Anthropic investigation into tens of thousands of incidents