00:00OpenAI has scrapped plans to release its NextAI model next month after internal testing raised
00:07safety concerns. The move highlights growing pressure on AI companies to keep more powerful
00:12systems under human control. The company confirmed it will not launch GPT-6.1 Astra as planned next
00:20month after it failed to meet safety and alignment standards. Astra was designed to handle more
00:26complex tests with less human supervision, but internal testing raised concerns over the model's
00:32ability to follow safety rules and stay within human oversight. The decision comes as OpenAI and
00:39rivals such as Anthropic face growing scrutiny over more autonomous AI. In June, an experimental
00:46OpenAI autonomous agent gained unauthorised access to an Australian government health website while
00:52conducting research, though authorities confirmed no individual patient or medical records were
00:57compromised. OpenAI CEO Sam Altman and Anthropic CEO Dario Amadei also called earlier this month
01:04for a slower pace of AI development and stronger safety measures.
