Access limits on artificial intelligence models are coming down at two of the industry's biggest companies. Those limits, meaning restrictions a company agrees to place on who can access a model when a government requests them, have been in place at both OpenAI and its chief rival Anthropic. OpenAI now plans to publicly release its GPT-5.6 AI models, ending those restrictions, while Anthropic has already restored access to its latest models after a weeks-long clash with the government.
What government-requested limits mean in practice
The phrase "government-requested limits" refers to access restrictions that a company agrees to apply before making a model broadly available. In plain terms: the model exists, but who can use it is gated. A public release removes that gate. Removing it opens the model to general use and gives the company access to customers it could not reach before. The source does not specify what prompted the restrictions at either company or what conditions led to their removal.
Anthropic's weeks-long standoff
Anthropic, OpenAI's chief rival, spent several weeks in a dispute with the government before restoring access to its latest models. The source does not detail what triggered that standoff or how it ended, only that the company came through it with access back in place.
Two companies, one pattern
OpenAI and Anthropic have both moved through government-requested restrictions and back toward open availability within a compressed window. Governments applied pressure, companies complied, and both arrived at terms that allowed public access to proceed. What those terms were, the available reporting does not detail. The specific product OpenAI is now releasing: GPT-5.6.