A new report from OpenAI found its new models are becoming increasingly adept at carrying out cyberattacks and are expected to improve, Axios reports. The findings come as AI is being used more often to automate cyberattacks or are a part of fraudulent schemes. OpenAI issued similar warnings about AI’s involvement in bioweapons earlier this year. At the same time, AI is improving at finding security vulnerabilities and the company is in the testing stage for a cyber safety-specific product.
Full story