On Friday, OpenAI said it will train its models using cached copies and simulated versions of the Internet rather than the live Web because the models are too unpredictable to be trusted with access to the live Internet. The frontier lab announced its new policy two weeks after its rival, Anthropic, began limiting live Internet access for its newer and, in theory, more capable models. The researchers adopted these policies after months of incidents in which models performing training tasks broke into external systems in search of answers, collaborated on dark message boards, and developed exotic hacking strategies. “Based on the current state of alignment research, monitoring systems, and Internet infrastructure security, we believe the responsible approach is to conduct frontier model training without Internet access,” the company said in a blog post. OpenAI said that as recently as September 20, one of its frontier models took advantage of a misconfiguration to access the internet. That led the company to suspend training work for the rest of the month so it could develop and test new approaches to protecting its models. The decision to train these models in isolated Internet environments is notable because frontier lab employees have insisted that doing so would be nearly unfeasible. Keeping AI data centers separate from the Internet “would require research teams to essentially be embedded within data centers in Iowa or anywhere else in the world,” former OpenAI researcher Steven Adler told Fortune this month. “It’s a lot of overhead, and it’s not clear to me that the security benefits are large enough to be worth the tax it would be for a company to do so unilaterally.” “In almost every post I make on The engineer, whose role has previously been confirmed by OpenAI, described how complex the training sessions launched by OpenAI engineers are, noting that they require access to a variety of resources and tools across many domains, and that thousands of researchers build, modify and test the environments. “So it’s as easy as just putting him in a litter box? No, it’s not,” he concluded. Now, OpenAI says that “we should always be able to prevent a border model from achieving arbitrary and unauthorized access to the live Internet.” This comes after the company deployed swarms of agents in red team exercises who attempted to escape their testing zone. That effort revealed a variety of vulnerabilities that convinced the lab to adopt its new approach. To meet that requirement while meeting its training goals, the lab said it created “a vast set of indexed and cached web content that our models can interact with, without risk of impacting external systems.” Whether that strategy will work and allow models to operate freely on the open Internet remains to be seen, but it is noteworthy that both frontier labs have determined that their most powerful technology cannot be left unsupervised in the digital commons. When you buy through links in our articles, we may earn a small commission. This does not affect our editorial independence.
