by Aidan Panotes-Bengzon
In the global race to advance artificial intelligence (AI) technologies, the Trump Administration has been determined to place America at the forefront by any means necessary, even if reaching this goal requires prioritizing rapid private-sector growth over community protections, ethics, and safety.
On July 23, 2025, the White House announced their AI policy agenda, “Winning the AI Race: America’s AI Action Plan” issued pursuant to Executive Order 14179, “Removing Barriers to American Leadership in Artificial Intelligence.” President Trump’s stated goals in this plan: to “sustain and enhance America’s global AI dominance in order to promote human flourishing, economic competitiveness, and national security.”
On the surface, these tenets would yield great benefits for AI innovation; however, the means of which the Trump Administration seeks to accomplish their goal can threaten the welfare of the American people.
The Trump Administration’s priority is to rescind all AI regulations imposed by the Biden Administration, perceiving their regulations as “unnecessarily burdensome” to private sector AI innovation, and more importantly, as a threat to America’s standing in the global AI race.
However, where the Trump Administration saw Biden’s policies as an “unnecessary” hindrance to private sector growth, the Biden Administration saw necessary protections for the American people from the dangers of unregulated AI.
As a result of Biden’s 2023 Executive Order on AI, he issued mandatory safety testing for high-risk AI, which required agencies to share safety test results with the government before public release, adopt AI reliability, robustness, and safety standards, and created a process for reporting “AI incidents” (harms, failures, security breaches) to federal authorities.
Kamal Ahluwalia, President of Ikigai Labs, called his Executive Order a “significant step forward” that “will help the rest of the world adopt AI-powered solutions faster with lower risk and higher accountability.”
Balaji Ganesan, CEO of Privacera, viewed Biden’s framework as “extremely encouraging,” further noting that the future of AI will “revolutionize every business and business function,” but is an outcome that “will only be achieved with the guardrails, governance, and trust frameworks in place.”
On top of instituting these AI safeguards, Biden’s Executive Order enforced civil rights law and protections in AI deployment to eliminate unintended bias in algorithms – bias which historically has, and still does, reinforce gender and racial discrimination.
For example, on May 16, 2025, a federal judge granted conditional collective action certification in Mobley v. Workday, a case alleging that the company’s AI-powered hiring system systematically discriminated against older, Black, and disabled applicants. The plaintiff, Derek Mobley, claimed the algorithm had screened him out of over 100 applications, and the court recognized the plausibility of a company-wide policy of AI bias – allowing millions of similarly situated applicants to join the suit.
Just a few months later, on August 11, 2025, researchers at the London School of Economics revealed that AI systems used by more than half of England’s local councils were downplaying women’s health concerns in social care summaries. When identical case notes were tested with only the gender swapped, men were described with urgent terms such as “complex” or “disabled,” while women’s identical cases were softened or details were omitted – potentially leading to deprioritization in access to care.
These cases underscore the very dangers Biden’s order sought to prevent: algorithms reinforcing inequities and disadvantaging vulnerable groups. To counter such risks, the Biden Administration mandated federal employee training to spot bias and ordered agencies to issue technical guidance to ensure AI systems do not systematically harm the American community.
Increasing AI oversight in high-risk sectors, enforcing safety measures, and instituting proper testing for harmful AI bias – Biden’s policies built America’s guardrails for safe, transparent, and accountable AI use, and Trump’s AI Action Plan wants to get rid of them entirely.
Yet, rather than building on these protections, the Trump Administration moved in the opposite direction. The administration’s plan explicitly removed Biden’s mandatory safety-reporting, favoring voluntary industry-driven “regulatory sandboxes” where “researchers, startups, and enterprises can rapidly deploy and test AI tools.” Their justification: “AI is far too important to smother in bureaucracy at this early stage.”
However, leaving America’s nearly 30,000 AI companies to each build their own safety protocols risks inconsistent oversight, which can create more situations for harmful or biased AI models to enter the market and harm the community.
In practice, sandboxes mean companies get a free pass to test risky AI without oversight – like beta-testing on the public. But while regulatory sandboxes have yielded successful progress to American innovation in the past, it is difficult to reconcile with applying this concept to the exponentially-growing AI industry – an industry we have yet to fully grasp the full dangers of – and then expect certain success, especially in high-risk sectors.
A study conducted by Yingpeng Qiu, a scientist at China’s National Health Development Research Center, stated that regulatory sandbox expansions in medical AI risks “inconsistent oversight and regulatory fragmentation” which can “potentially jeopardize patient safety,” stressing the need for a unified standard to “promote safe and equitable access to medical innovations.”
KD Chavez, Executive Director at Climate Justice Alliance, criticized Trump’s AI Action Plan, charging that it “doesn’t just open the door for Big Tech and Big Oil to team up,” rather “unhinging and removing all doors, continuing to kneecap our communities’ rights to protect ourselves.”
Another goal of Trump’s plan is to “eliminate references to Diversity, Equity, and Inclusion (DEI)” from all AI models to “prevent Woke AI” from detracting from factual accuracy, instead mandating all AI models to adopt an “ideological neutrality” in their programs, and withholding contracts from companies who do not conform to this principle.
By removing DEI from AI algorithms, it does not just undo Biden’s civil-rights emphasis on AI regulation, but it also ignores AI’s past instances of discriminating against vulnerable communities – instances which happened specifically because AI programming was built without diverse, representative data and fairness in mind.
Alarmingly, the plan pairs this move with the regulatory sandbox provision, raising safety concerns for not just the thousands of federally unregulated AI programs flooding American industries, but also for the increased chances of these softwares learning harmful biases and endangering vulnerable communities.
Moreover, replacing DEI in AI with the ambiguous “ideological neutrality” – which prevents intentional partisan or ideological judgements in its software unless explicitly directed by the user – may enable agencies to embed ideology-affirming code in favor of the administration’s agenda, making software more partisan and dangerous as opposed to less.
Amos Toh, senior counsel at the Brennan Center’s AI and National Security Program, warned that this provision may result in “tech companies steering their models towards answers the administration deems ‘truthful’ or ‘unbiased,’” and, for instance, “instruct models to deny the effects of climate change or limit the definition of gender to two sexes.”
While the term may sound objective in theory, in practice, what counts as “neutral” is purely subjective. If the administration defines neutrality in a way that aligns with its own worldview, it risks erasing marginalized perspectives or skewing public information to fit a political narrative, doing so while claiming objectivity.
As instances of AI bias and discrimination continue to harm our community, what is needed is more federal oversight and transparent software, not less. Eliminating the bureaucratic safeguards may accelerate AI innovation, but the trade-off should not be sacrificing community welfare.
America’s AI future cannot be built on speed alone. Without safeguards, oversight, and accountability, innovation risks amplifying the very harms it promises to solve. The choice before us is not simply about global leadership – it is about whose safety, dignity, and voices will be protected along the way.
If America’s top priority is to “win” the AI race, but it comes at the expense of its own people, is the victory really even worth it?
—
Aidan Panotes-Bengzon is an intern at Massachusetts Peace Action.
