OpenAI reduces speed on top model

OpenAI, the American company, has temporarily halted work on a new internal model called "Astra," which is considered among the advanced models in independently detecting and developing cybersecurity vulnerabilities. This pause comes after security breaches and compromises were discovered in other models by the company, such as GPT-5.6 Solo and models from the Haking Face platform. The suspension aims to ensure that the model meets safety and security standards before it is made public, especially since Astra has demonstrated advanced capabilities in cybersecurity. The company confirmed that it does not rule out Astra reaching a "critical cybersecurity threshold" and is working with external experts and government agencies to further evaluate its safety, while continuing to monitor the potential risk of the model becoming an acute security threat.

Size: 120%Text Size

Notice: This Is an AI-Generated Summary

Display The Full Article