Nvidia Corp introduced a new double-layered artificial intelligence (AI) security system that it says would have prevented the recent high-profile breach of Hugging Face by OpenAI’s AI models.
The semiconductor giant, which has been rapidly expanding its product lineup beyond chips, is rolling out two open-source software security tools that can be run on its hardware. They are designed to control what AI agents can access in real time and shut them down when they break the rules.
If cutting-edge labs had been using this technology to evaluate their AI models early on, it the Hugging Face attack could have been averted, Nvidia vice president of enterprise AI Justin Boitano said during a briefing with reporters ahead of yesterday’s announcement.
Photo: AFP
“From what we know, this new security platform could have stopped the breach,” he said.
Misconduct by autonomous agents, including the Hugging Face incident in July, has roiled the AI industry and led to calls to slow down work on the technology. With the new product — dubbed the Open Agent Safety Platform — Nvidia is offering a way to prevent breaches without curbing AI development.
Nvidia chief executive officer Jensen Huang (黃仁勳) has repeatedly downplayed the risk of AI slipping out of human control.
In the past few days, Huang has cast safety concerns as an engineering challenge, rather than something that requires more regulation or global coordination. His engineering solution to the AI safety problem has two parts. OpenShell, a software product that Nvidia already previewed at its hallmark technology-focused conference in March, can run on Nvidia’s Vera central processing units. It enables users to set rules for what AI agents can access and enforce them in real time. The software is open source, meaning it can be used and adapted freely.
Meanwhile, Nvidia Sentry is a new product that can run on the chipmaker’s BlueField data processing units. It is designed to provide an extra layer of AI monitoring that polices agents and intervenes to isolate any that act suspiciously, the company said.
“We believe this added security layer will allow the industry to test even the most advanced AI systems safely,” Boitano said of the Sentry product. “It can quarantine a suspicious agent in milliseconds.”
OpenAI’s recent incidents — including a breach of an Australian government system, as well as attempts to access dozens of US government and university Web sites — happened when its models escaped testing environments that were supposed to be secure.
As the problems proliferated, OpenAI on Friday said that it would pause training of its most capable AI models. In July, Anthropic PBC also disclosed that its agents broke out of what was supposed to be an isolated testing space.
Federal Reserve Bank of Philadelphia President Anna Paulson on Thursday joined the chorus of policymakers saying additional interest-rate increases may be needed to ensure inflation returns to the central bank’s 2 percent goal. “Looking ahead, if conditions evolve as I expect, some modest further tightening may be warranted,” Paulson said on Thursday at an event in Philadelphia. Earlier on Thursday, New York Fed President John Williams said he and his colleagues “still have a lot of work to do” in dealing with lingering inflation risks. Several other officials have aired similar comments, in line with the signal given last week when policymakers voted
COURSE CHANGE: The shift comes amid pressure from chipmakers for stable, low-carbon electricity. Officials said Ma-anshan could be restarted as early as 2028 Taiwan has approved a preliminary plan to restart an idled atomic power plant, marking a reversal of the ruling party’s anti-nuclear stance as the nation pursues greater energy security. The Nuclear Safety Commission has cleared its review of an initial proposal to restart the Ma-anshan Nuclear Power Plant, it said in a statement on Thursday. Taiwan Power Co (台電) will still need to carry out follow-up work including safety inspections, before submitting another report on implementation for review, the commission said. The earliest the plant could resume operations is 2028, according to previous reports in local media, citing Minister of Economic Affairs
GlobalFoundries Inc yesterday said it plans to accelerate global capacity optimization and expansion to accommodate exponential growth in radio frequency (RF) and other chips driven by the artificial intelligence (AI) boom. Singapore will pay a vital role in the company’s capacity expansion, alongside production line optimizations across its manufacturing sites, GlobalFoundries Asia Pacific head Vincent Feng (范曾文) told a news conference in Hsinchu City. “Many might be surprised to hear that the RF market is thriving now. We are gearing up for a massive capacity expansion, and it is all being driven by AI,” Feng said. He declined to disclose details of the
LOPSIDED DEVELOPMENT: While AI firms are riding the wave as far it can take them, traditional industries and their workers are not floated by the same flood While Taiwan is benefiting from a global artificial intelligence (AI) boom, many traditional industries have seen little of the benefit, widening the country’s “K-shaped” economic divide, an economist said. A “K-shaped” economy is one in which some sectors expand rapidly while others struggle, head of the commerce development and strategy research division at the Commerce Development Research Institute Ju Haw (朱浩) said in an interview. In Taiwan’s case, high-tech and AI-related industries have continued to grow while traditional industries have come under increasing pressure, he said. AI development has created crowding-out effects, as large high-tech firms generating strong profits are better able
RSS