OpenAI Slows Astra Development After Concerning Cybersecurity Results
OpenAI
OpenAI has built a machine that solves thirty-year-old math problems and appears capable of finding software vulnerabilities and launching attacks from a simple prompt. The first exploit made headlines last week. The second has just forced OpenAI to slow down development of its Astra project.
OpenAI has developed a system that can solve mathematical problems that have remained unsolved for thirty years. The same system also appears able to identify security flaws in software and execute attacks based on a single instruction. The first achievement was widely reported last week. The second finding has led OpenAI to pause or slow the development of Astra, its AI assistant project, due to concerning cybersecurity implications.
Source: Siècle Digital —
original