A source familiar with the negotiations revealed that before OpenAI had numerous cybersecurity incidents related to its technology and serious warnings from industry practitioners, OpenAI had already negotiated a legally binding agreement with Anthropic, and the two sides planned to stress test each other's big models. The source said that earlier this year, the two companies and their legal teams have been finalizing the details of the agreement and are planning to conduct various tests on their respective models to investigate loopholes and potential hidden risks. It's unclear whether the two parties signed the agreement before OpenAI experienced multiple incidents and unreleased models invaded its own and other enterprise systems. A person familiar with OpenAI's strategy said that the company has been discussing various solutions to explore cooperation models between enterprises and between enterprises and the government in the field of AI security. The source said that discussions have recently expanded to a new safety mechanism, covering the two major stages before model training and before official release.

Disclaimer:Webull uses external vendor Google Translation Service for news translations where we endeavour to ensure these are correct, however, we recommend that you please double-check this information accordingly. Webull is not responsible for translation errors or issues.