Defense officials Steve Feinberg, founder of Cerberus Capital Management, and Emil Michael, formerly Uber’s second-highest executive, were reportedly present at a contentious meeting with Anthropic at the Pentagon.
This edition of Regulator, The Verge's exclusive newsletter for subscribers, delves into the influential figures, prominent personalities, and evolving artificial intelligence models navigating the corridors of power in Washington. Readers not yet subscribed are encouraged to join.
In significant news, Regulator is launching a new tip line, inviting readers to submit commentary, valuable insights, and confidential information for potential publication via [email protected].
Earlier today, a report was published detailing the Pentagon's rigorous contract renegotiations with Anthropic, preceding a crucial meeting between Defense Secretary Pete Hegseth and Anthropic CEO Dario Amodei. The negotiations carry considerable weight, as the Pentagon has declared Anthropic a "supply-chain risk" should it fail to align with the department's stipulations regarding its acceptable use policy.
Following the meeting, Axios reported that Secretary Hegseth had included several senior Defense officials, a move intended to underscore the Pentagon's serious approach to the dispute. However, within the context of a "post-DOGE Trump administration run by broligarchs," scrutinizing attendees' backgrounds is deemed prudent. While some participants were career government and military personnel, others presented notably unconventional professional histories:
The attendance of Feinberg and Michael warrants attention. Both possess relevant defense sector experience: Michael served as a White House fellow during the Obama administration and spent two years as a special assistant to Defense Secretary Robert Gates at the Pentagon. Feinberg, too, has a clear history with defense contracts. Nevertheless, it is crucial to acknowledge the aggressive business ethos that private sector individuals often introduce into government roles, particularly during critical negotiations of this nature. Parnell's attendance, conversely, is logical given his role as spokesman for Pete Hegseth.
A critical aspect of these negotiations, which was not extensively covered previously, is the emerging issue of "single-supplier vulnerability."
In 2024, the Biden administration issued a national security memorandum on artificial intelligence, establishing directives for supply chain protection. A key mandate required the Department of Defense to engage with at least two frontier AI laboratories approved for classified information, thereby mitigating the risk of a single vendor compromise collapsing an entire IT infrastructure. However, prior efforts by the Trump administration had already sought to address this vulnerability. Despite having separate contracts with Anthropic, Google, xAI, and OpenAI, only Anthropic's model was cleared for classified use when Hegseth released his new AI policy memo in January.
This situation has created a dilemma for the Pentagon: even if Anthropic were successfully disengaged, and defense contractors underwent the arduous process of removing Claude from their operations, the department would risk non-compliance with its own established guidelines, in addition to disregarding a fundamental principle of tech industry best practice – avoiding single-supplier vulnerability.
This predicament sheds light on the Pentagon's recent decision to unexpectedly grant xAI’s Grok access to classified systems, despite its general perception as the least capable of the currently available models. The New York Times has reported that Google is nearing an agreement to permit the Pentagon to utilize Gemini for classified operations, with defense experts considering Gemini a strong competitor to Claude. In contrast, xAI’s Grok is "not considered as advanced or as reliable as Anthropic’s." OpenAI, however, is not close to securing a deal, reportedly due to the company's belief that ChatGPT’s safety features require further enhancement before deployment on classified networks.
Considering the requirements, the Pentagon must engage with two out of four available AI models. The options presented are:
1) An AI provider offering a competent model, but exhibiting evolving ethical stances.
2) The provider of the superior AI model, which, however, prohibits its use for autonomous lethal operations without human oversight.
3) A company whose AI model currently lacks the requisite security for deployment.
4) A company whose AI has demonstrated concerning issues such as racist hallucinations, the generation of illicit content, and is not deemed "advanced [or] reliable."
Should contracts with options 2 (the superior, ethically constrained model) and 3 (the presently insecure model) prove unfeasible, the Pentagon would be left with options 1 (the competent but ethically adaptable model) and 4 (the problematic and unreliable model). This outcome, as even defense officials concede, is far from ideal for national security. A Defense official, speaking to Axios before the meeting, emphasized the dilemma: “The only reason we’re still talking to these people [Anthropic] is we need them and we need them now. The problem for these guys is they are that good.”
Last week's Clarity Act negotiations, involving finance and crypto sectors, unexpectedly evolved into another iteration of a recurring phenomenon: Laura Loomer's public commentary on niche technological matters, framed as litmus tests for MAGA allegiance.
On Thursday, a select group of influential figures from the cryptocurrency and finance industries convened at the White House to further refine draft language pertaining to stablecoin yields. Coinbase, whose withdrawal of support from Clarity over stablecoin yields had initiated these negotiations, was present. However, preceding this meeting, Loomer issued a characteristic tweet, employing her typical strategy to sway Trump: portraying the target as having previously supported Trump’s adversaries, thereby branding them as disloyal.
Ironically, Coinbase has since emerged as a prominent corporate advocate for the Trump administration, contributing financially to its key initiatives and even featuring its logo conspicuously at last year’s military parade.
Despite Loomer having expressed similar sentiments regarding Coinbase last June, her recent tweet appears to have had no discernible effect on Coinbase’s access to Trump, nor is it likely to in the near future. It has been reported that CEO Brian Armstrong was present at Mar-a-Lago the day prior to Loomer’s tweet, attending a World Liberty Financial event.
For those who followed Logan Paul’s Pokémon card collection auction, it was revealed last week that one card fetched a record-breaking $16.5 million. The identity of this significant purchaser is AJ Scaramucci, son of the notable New York financier and former Trump ally, Anthony Scaramucci, who famously served as Trump’s White House Communications Director for a brief ten-day period in 2017.
AJ Scaramucci, founder of Solari Capital, previously invested $100 million in a Bitcoin mining platform managed by Eric Trump. He is now the owner of the highly coveted Pikachu Illustrator card – one of only 39 in existence and in pristine Grade 10 condition – along with the diamond chain and display case Paul utilized during his WrestleMania 38 appearance. Scaramucci informed reporters that this acquisition is part of an ambitious "planetary treasure hunt," expressing aspirations to also acquire a T. rex skull and the Declaration of Independence. He later stated on X (formerly Twitter) his intention to house the card at the Nintendo Museum in Kyoto, solidifying its status as "the ‘Mona Lisa’ of the Pokemon franchise."
It is astounding that a court, specifically the Southern District of New York, found it necessary to clarify that placing correspondence between a client and their attorney into a publicly accessible AI platform nullifies attorney-client privilege, rendering such communications subject to discovery.
On that note, we wish you a pleasant State of the Union watch party, for those who partake, and look forward to connecting again next week.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.