OpenAI has tightened controls around its unreleased Astra model after preliminary testing showed it could potentially reach a critical level of cyber capability.
The company said it could not rule out Astra being capable of launching sophisticated cyberattacks autonomously. OpenAI has halted some internal activities involving the model and introduced stricter safeguards, including isolated testing environments, expanded monitoring and detection systems.
US lawmakers are also pushing the AI Kill Switch Act, which would require companies to maintain systems capable of shutting down or suspending advanced models.
According to the report, regulators in the US and European Union are also developing stronger oversight measures as AI capabilities advance rapidly.
Related Tweet:
OpenAI tightens controls on its new model over cybersecurity risks, as AI security debate intensifies https://t.co/Yd0HQszlEi
— CNBC Tech (@CNBCtech) August 10, 2026
Also Read:
TIPP Staff
