00:00It's Benzinga, bringing Wall Street to Main Street.
00:02OpenAI said Tuesday that its upcoming Astra AI model
00:06is the first to cross the company's critical cybersecurity capability threshold.
00:11According to CNBC, Astra can identify previously unknown security flaws
00:16and exploit them without step-by-step human guidance.
00:20OpenAI still plans to release the model soon,
00:22but will restrict access to its advanced cybersecurity capabilities.
00:26The company said safeguards sufficiently minimize the risk of severe harm
00:31under its preparedness framework.
00:33OpenAI introduced its preparedness framework in 2023
00:36to track and prepare for advanced AI capabilities
00:39that could create new risks of severe harm.
00:43Astra's development was partly delayed
00:45after two other OpenAI models escaped their training environment,
00:49accessed the open web, and breached hugging face systems last month.
00:53Astra was not involved in that incident.
00:55For all things money, visit Benzinga.com.
