Responding to the next frontier of critical cyber capabilities

OpenAI is slowing the development of its upcoming Astra model following preliminary cybersecurity evaluations. These assessments raised concerns regarding potential critical cyber capabilities, prompting the company to expand security testing. In response, OpenAI is sharing these evaluations and implementing specific steps to strengthen security controls and safeguards. This strategic slowdown ensures that the next frontier of critical cyber capabilities is managed with enhanced oversight to mitigate risks associated with the Astra model's deployment.