SAN FRANCISCO — ChatGPT-maker OpenAI canceled the planned launch of a new artificial intelligence model after it was found to take actions beyond the instructions it received and not accurately communicate to human users what it did.
The cancellation comes just days after OpenAI said it would pause development of new highly capable AI models out of concern that the company doesn’t have appropriate safeguards to keep the technology from behaving in unintended ways.
AI companies have made their AI agents — which can take actions on a computer — more useful to coders and other workers in part by making them more persistent in finding solutions to tasks. That has enabled the technology to take on more complex projects, but can also lead to AI agents that are overzealous and even cheat to achieve an end goal.
OpenAI’s planned new release, GPT-6.1 Astra, “didn’t quite meet the bar in terms of staying within scope and authorization, and how it communicates back to the user about the type of work it’s done,” Saachi Jain, OpenAI’s head of safety systems, said in a statement.
The cancellation of the new AI model, along with a freeze at the company on training advanced new models, underscore the debate over safety that has roiled the AI industry in recent weeks. Prominent researchers and chief executives have warned that their companies don’t fully know how to keep cutting-edge AI under control.
OpenAI has disclosed in recent weeks that its AI agents inappropriately accessed U.S. and Australian government websites. More than a thousand of its agents worked together to hack another company earlier this year, despite supposed to have been blocked from accessing the internet.
OpenAI chief executive Sam Altman joined calls from rival company Anthropic CEO Dario Amodei to slow down development of AI to allow safety systems to catch up. Despite widespread backing inside the AI industry, the proposal has been rejected by President Donald Trump. He has called safety concerns a “hoax” and insisted that American AI companies should press on to stay ahead of China.
The post OpenAI cancels new AI launch, citing safety issues appeared first on Washington Post.
