OpenAI delays latest model over security concerns, as industry faces new safety pressures

The company attributed the decision to security concerns voiced by its own researchers. The excerpt does not name the researchers, describe the specific concerns, or state how long the release will be postponed. The AP headline describes Astra as OpenAI's latest model and refers to new safety pressures facing the industry.

The announcement came one day before AI executives are set to meet with President Donald Trump in Washington, the article states. According to the report, technology companies are facing new pressure to be accountable for how their models can be abused. The excerpt does not identify which executives are expected to attend, where the meeting will be held, or what is on its agenda. It also does not state whether OpenAI's decision was connected to the meeting.

The article places the delay in the context of that broader accountability pressure on AI developers. It does not state whether Astra has been completed, what capabilities it would include, or when OpenAI had previously intended to release it. The report includes no comment from OpenAI beyond the stated reason of concerns raised by its researchers. It also does not indicate whether other AI companies have made similar decisions about their own models, and it does not say whether any technical documentation was published alongside the announcement.