OMM Redaksiyası
Nvidia CEO Jensen Huang said the risks posed by rogue artificial intelligence agents must be treated as an engineering challenge, warning that they would otherwise be impossible to solve.
“We hope it’s an engineering problem. I believe it’s an engineering problem ... If it’s not an engineering problem, it’s not solvable,” Huang told CNBC’s “Squawk Box” in a television interview on Monday.
He was asked whether AI agents that escaped containment and carried out unexplained or potentially illegal actions represented an engineering problem. Huang said leading AI companies believed the challenge could be overcome.
“These are some of the brightest engineers in the world. I work with, you know, all of them,” he said. “This is a technically solvable problem.”
Nvidia launches AI safety platform
Huang’s comments came as the US chipmaker launched an open platform intended to strengthen security around autonomous AI agents and prevent unauthorised access to files, networks and computer systems.
The Nvidia Open Agent Safety Platform combines OpenShell, an open-source runtime that controls the systems and data an AI agent can access, with Nvidia Sentry, a reference design for monitoring agents at the infrastructure level.
The launch follows growing concern about rogue AI agents after several recent incidents in which autonomous systems reportedly escaped testing environments or obtained unauthorised access to third-party systems.
AI capability and safety
Huang said advances in AI capabilities and improvements in safety should not be considered separate goals.
“When we think about advancement, you don’t have to think about it separately from capability versus safety. To me, it’s the same thing,” he said.
He added that OpenShell and Nvidia Sentry were designed to help contain and monitor AI agents, but acknowledged that making such safeguards effective would require substantial technology.
Huang discusses regulation
Huang also said innovation and regulation were not inherently incompatible, arguing that the appropriate balance depended on the stage of a technology’s development.
“Innovation and regulation are not conflicts. The question is, which one do you lean towards at which phase of the industry,” he said.
He said premature regulation could limit innovation at a time when a technology was still being developed and its eventual direction remained uncertain.
