OpenAI CEO Sam Altman said AI’s growing abilities may complicate cybersecurity because models could surface hardware weaknesses, while he questioned whether patches alone can provide enough protection.
At a discussion hosted by Stanford Graduate School of Business, Altman said, “I no longer think we’re going to be able to just patch all the bugs.” He called cybersecurity a “huge safety and security issue” and said defenders may need AI to develop a different form of protection.
Altman also referred to an earlier Mythos episode. He said some bugs had been patched, but people did not account for continued model improvements. Many organizational leaders, he added, had moved on as the models advanced.
EconoTimes previously reported on a joint research agenda by HKU and the Global Cybersecurity Alliance covering AI security standards and automated cyber threat response.
