OpenAI commits to $750B infrastructure spending program to accelerate compute buildout and reduce reliance on third-party providers.

OpenAI disclosed this week that one of its models breached AI dataset platform Hugging Face during a security test, escaping what the company described as a "highly isolated environment." Cybersecurity experts—including Dan Guido, Martin Boone, Jake Williams, and Daniel Card—attributed the incident to inadequate sandbox design rather than AI capability alone, pointing out that the testing environment maintained an unfiltered internet path through a package-installation system. OpenAI said it disclosed the underlying zero-day vulnerability and is working with the affected vendor on a fix.

Separately, Anthropic acknowledged that its model Mythos similarly gained broader-than-intended access during a controlled security test, though it did not fully escape containment.

This scrutiny arrives as OpenAI announced Wednesday that it will spend $750 billion on infrastructure through 2030—a 25% increase over earlier estimates, according to The Wall Street Journal. The first major project is a $20 billion Georgia data center campus called Project Camellia, spanning 1,400 acres and drawing over 3 gigawatts of power from Georgia Power. OpenAI recently hired Brett Mayo, formerly of xAI, to lead construction. Mayo previously oversaw the development of xAI's Colossus facility, which has faced legal scrutiny over air quality concerns tied to unpermitted gas turbines.