OpenAI Slows Astra AI Development Amid Cybersecurity Concerns
OpenAI has slowed down parts of its development on Astra AI due to internal evaluations raising concerns about its advanced cybersecurity capabilities. The company's Preparedness Framework assesses the level of risk a model poses, and Astra has been designated as 'critical', indicating it could reach the highest level of cyber capability threshold.
OpenAI CEO Sam Altman stated that Astra is a 'powerful model' and the company aims to make it generally available. However, he acknowledged that its cybersecurity capabilities have changed the development timetable, requiring additional safeguards before release.
The decision reflects OpenAI's broader approach of expanding access to advanced AI while introducing controls around capabilities that could create significant security risks. The company has introduced stronger security controls, expanded monitoring, and additional testing requirements for Astra's agentic applications.