OpenAI postpones launch of ‘GPT-6.1 Astra’ over security concerns

OpenAI has announced the postponement of the launch of its new artificial intelligence model, dubbed “GPT-6.1 Astra,” following security concerns raised by its researchers. The move is aimed at slowing the pace of development to ensure that the technology meets safety standards for increasingly autonomous systems.

According to The Wall Street Journal, the announcement comes on the eve of a meeting expected to bring together artificial intelligence industry executives and U.S. President Donald Trump in Washington, amid growing pressure on technology companies to take greater responsibility for unauthorized uses of their models.

In this regard, Sashi Jain, OpenAI’s head of safety systems, said in a statement that the upcoming release had not yet fully met the required standards. She noted that the model had demonstrated a greater ability to complete tasks, while stressing the need to balance these capabilities against the risks of unauthorized behavior.

Jain reaffirmed the company’s commitment to ensuring the model’s safety and alignment, both during internal testing and after its eventual release, highlighting the adoption of high standards for safety and alignment.

OpenAI had also decided last week to suspend the training of its most advanced models, saying work would resume once sufficient additional safeguards had been put in place.