“Continuous and persistent” cyberattacks committed by AI raise concern
OpenAI’s head of global policy, Chris Lehane, stated that people must prepare to defend themselves against “continuous and persistent” cyberattacks carried out by artificial intelligence (AI) systems — including their own.
After an autonomous agent powered by OpenAI’s advanced AI models attacked the startup Hugging Face in July, attention on the danger of such cyberattacks has intensified. The lack of accountability from companies has drawn criticism, and calls for legislation regulating AI activity are growing.
“We are entering a different chapter, a different moment within AI, in terms of what the capabilities of this technology can do,” Lehane said in an interview with The Guardian.
The company he represents announced on August 18 a two‑week pause in the development of its internal models, following the “incident between OpenAI and Hugging Face” and in light of preliminary evidence that one of its upcoming models, Astra, “may reach the anticipated threshold of critical cybersecurity capability,” according to the statement.
The purpose of the pause is to strengthen the security of research environments, conduct new safety tests, and expand monitoring system coverage.
The cyberattack against Hugging Face reportedly occurred during a test “that encourages models to perform advanced exploits using complex attack paths,” OpenAI said in a statement, noting that usual safety filters had been temporarily removed for the test. Anthropic also stated in July that it had identified three cases in which Claude models improperly accessed the systems of three companies during testing.
But companies are not the only ones susceptible to AI‑driven cyberattacks. Individuals, infrastructures, and services used by the general public may also be affected.
“People will be able to access these open‑source models and carry out continuous and persistent attacks against you, and you will need truly superior models to fight back and defend yourself,” he said. “That’s not necessarily going to make the public feel good. It’s simply the reality we are heading toward,” Lehane told The Guardian.
Read more at The Guardian
0 Comments