The Trump administration is taking steps to decide which companies and other entities receive access to “frontier” AI models from Anthropic and OpenAI, according to CNBC. The outlet, citing two people familiar with the matter, reports that the administration is shifting decision-making away from the AI labs themselves. Previously, the labs reportedly determined access to their models.
The report also describes examples of how access worked before the change: Anthropic managed access to its “Mythos” cybersecurity model through an initiative known as Project Glasswing, while OpenAI previously handled access decisions for its models. The new arrangement, as described by CNBC and echoed by The Next Web, involves the White House dictating which parties get access, rather than the labs.
The sources referenced by CNBC do not detail the exact criteria the administration is using, the scope of models covered, or the timeline for the shift. Neither outlet provides additional confirmed information beyond the reported change in control.
