As I’ve written before, this is one reason AI news is so confusing It’s not clear what AI companies are and aren’t allowed to do. But a voluntary deal with a Greater AI is reportedly in the works (Your mileage may vary on whether that’s a good thing).
According to the Financial Times“as soon as next week” the Trump Administration and several major US border AI companies are expected to announce a set of standards for border AI models, particularly related to cybersecurity capabilities. The report cites “people familiar with the talks” — anonymous leakers, in other words.
One of the FT’s anonymous sources said the Center for AI Standards and Innovation (CAISI), under the Commerce Department, and the National Security Agency (NSA), under the Pentagon, will be central to these standards.
On June 12, the U.S. issued an export control directive to Anthropic, essentially shutting down its most recently publicly released model and keeping it offline for the rest of June. OpenAI is apparently concerned that something similar might happen, and is scrambling its plans as well discontinued its latest modelsostensibly as a backup.
At the very beginning of the Trump 2.0 Administration, Vice President JD Vance hinted A laissez-faire approach to AI regulation. That has now changed significantly with the White House’s actions against Anthropic, the executive order on artificial intelligence, and these standards, which now appear to be formalizing certain aspects of the order.
Government, according to the order means:
“… develop and maintain a classified benchmarking process to assess the advanced cyber capabilities of AI models and determine the threshold at which an AI model should be designated a “covert boundary model” for the purposes of this order, sharing such assessments with AI developers and researchers, as appropriate.”
If the benchmarking process is indeed classified, it means the public won’t know what standards Big AI is held to. However, shared security practices among many companies will make it easier to achieve at least some of the agreed upon standards.
It is not entirely clear which companies will be parties to this voluntary inspection agreement. The FT article mentions Anthropic, OpenAI, Amazon, Microsoft and Google. Interestingly, there is no mention of Meta, and about a week ago, other anonymous sources familiar with these negotiations It has reportedly leaked that Meta is a stopgap and that the Trump Administration is working overtime to buy Meta.
