Trump administration’s AI plan takes aim at ‘anti-woke’ large language models

The White House released long-awaited AI guidance in July, and it’s a mixed bag. While the Trump administration’s “anti-woke” executive order takes aim at perceived diversity, equity, and inclusion (DEI) initiatives in the development of large language models, IT pros who spoke with IT Brew were pleasantly surprised to find a focus on open source in its separate AI action plan.

According to the EO signed by President Trump, DEI in regard to AI is considered the “suppression or distortion” of factual information about race or sex. It continues: “DEI displaces the commitment to truth in favor of preferred outcomes and, as recent history illustrates, poses an existential threat to reliable AI.” The Associated Press described the order as an attempt at censorship, and noted that it “still faces a study period before it gets into official procurement rules.”

Anti-woke AI. Asad Ramzanali, the current director of AI and tech policy at the Vanderbilt Policy Accelerator and former official at the White House Office of Science and Technology Policy under President Biden, told IT Brew that he imagines the implementation of the “anti-woke” order may look like a government-established test for systems.

“I worry that in the hope of creating neutrality, you just mandated an ideological purity test,” Ramzanali said. “I’m allowed to have a political view, you’re allowed to have a political view, your company can have a political view. That’s really tricky to push companies ostensibly through procurement to meet some kind of neutrality.”

While experts maintain that the anti-woke EO that follows the action plan’s release is concerning, they also agree that the administration’s positioning toward open-source models is a benefit to the private sector.

Open up! Seemingly more substantive guidance on AI is outlined in the administration’s action plan, titled “Winning the Race,” which points to the government’s ability to “accelerate the maturation of a healthy financial market for compute” through collaborating with industry and using other vehicles in the federal space, like the National AI Research Resource Pilot and the National Institute of Standards and Technology, for open-source models.

The administration recommends ensuring access to “large-scale computing power for startups and academics by improving the financial market for compute. Currently, a company seeking to use large-scale compute often must sign long-term contracts with hyperscalers—far beyond the budgetary reach of most academics and many startups.”

The action plan further states that the US has “solved this problem before” through spot markets (when assets are purchased or sold immediately without a long-term commitment) and forward markets (where contracts are made to purchase or sell assets at a determined price in the future). This is in concert with the industry and government approach the administration believes can hasten the maturation of a healthy market for compute.

It’s so over, and we’re so back? In addition to AI innovation, the action plan focuses on infrastructure and diplomacy—pointing directly toward encouraging both open-source and open-weight models.

Google’s Open Source webpage points to releasing open-source software as a benefit from the standpoint of having more perspectives. It states that open-source work has “benevolent results” but it is not “an act of charity.” It goes on to say that having an open-source approach can yield higher returns than the initial investment.

Cobalt, however, pointed to open-source software as having limited support and functionality, compatibility issues, and more. It looks to open-source security risks, as well, if the software is maintained incorrectly—specifically, vulnerabilities can allow attackers to access sensitive information or take control of a system.

Policy and industry experts, like Travis Hall, the director for state engagement at the nonprofit and nonpartisan Center for Democracy and Technology and Dhruv Patel, CEO of Syncurrent—a company that provides digital services to match local governments to grants—point to the action plan’s stance on open source as a positive for the market.

Hall highlighted the open-source and open-weight systems of the action plan as a potential positive.

“I think that this is a big deal, particularly for industry and IT professionals, because there have been debates…on whether or not open systems should be allowed, whether they can be exported, whether there should be expert controls on model weights or open-source systems,” Hall said. “It means that the tendency towards large model, proprietary systems being the only thing that is available is not going to be the case.”

Hall believes this aspect of the plan is a “pro-competitive move that will help IT professionals in particular” through having a broader set of choices than what is currently available—like those provided by Linux, WordPress, and Android. He also said that IT pros can benefit from having the ability to use open-source systems and “innovate on their own.”

“Everybody knows that open-source software is available, it is a thing,” Hall said. “It hasn’t made commercial proprietary software go away, but it does help to diversify the ecosystem and also help support a more competitive ecosystem.”

Patel told IT Brew that those who adopt open-source models will be impacted, but said he doesn’t know if it’s “for the better or for the worse.”

“Right now, I lean more good than bad, particularly because of the emphasis on supporting open source and supporting open models, as opposed to not,” Patel said.

He continued: “For tech businesses, for tech startups, it’s incredibly important that we have robust open-source models. But even more so for nutritional IT tech-enabled businesses, businesses that don’t directly have a fingerprint in tech but they’re tech-enabled.”