Brandon Fletcher
The current landscape of AI news is chaotic, primarily due to the unclear regulations governing what AI companies can and cannot do. a potential voluntary agreement among major AI players is reportedly being developed, which could bring much-needed clarity (opinions may vary on whether this will be beneficial).
According to the Financial Times, the Trump Administration, along with several leading U.S. AI companies, is poised to announce a set of standards for frontier AI models as early as next week. This announcement will particularly focus on cybersecurity capabilities. The report references “people familiar with the talks,” indicating reliance on anonymous sources.
One source cited by the FT mentioned that the Center for AI Standards and Innovation (CAISI), under the Commerce Department, and the National Security Agency (NSA), part of the Pentagon, will play key roles in establishing these standards once they are formalized.
On June 12, the U.S. issued an export control directive to Anthropic, effectively disabling its most recent publicly available model and keeping it offline for the remainder of June. Concerned about potential similar actions, OpenAI has delayed the release of its newest models as a precautionary measure.
At the outset of the Trump 2.0 Administration, Vice President J.D. Vance expressed a hands-off stance regarding AI regulation. This approach has shifted significantly, as evidenced by the White House’s recent actions against Anthropic, the executive order concerning AI, and the forthcoming standards that appear to formalize certain elements of that order.
According to the executive order, the government is tasked with:
“…developing and maintaining a classified benchmarking process to evaluate the advanced cyber capabilities of AI models and determining the threshold at which an AI model qualifies as a ‘covered frontier model’ for the purposes of this order, sharing such assessments with AI developers and researchers as appropriate.”
If the benchmarking process is classified, the public will remain unaware of the specific standards that Big AI must adhere to. Nevertheless, the establishment of shared practices and safeguards across various companies will likely provide insights into the agreed-upon standards.
It remains uncertain which companies will participate in this voluntary vetting agreement. The FT article mentions Anthropic, OpenAI, Amazon, Microsoft, and Google as potential participants. Notably absent from this list is Meta, which, according to reports from other anonymous sources, has been hesitant to join the agreement. The Trump Administration is reportedly making significant efforts to secure Meta’s involvement.
Daniel Mercer
David Bridges
