Of CEOs and Secretaries of War: A Call for a Whole-of-Society Approach
“The real problem is not whether machines think but whether men do.”
—B.F. Skinner
“We must learn to sit down together and talk about a little culture …”
—Sylvia Wynter
The ongoing tension between Anthropic and the Pentagon encapsulates the rising discussions surrounding AI ethics. This conflict highlights the critical issue of who should steer the design, development, and deployment of such a powerful technology, especially regarding its intended or unintended applications. At the core, it appears essential that elected officials—not a select group of corporate leaders—determine the framework within which influential technologies like AI are utilized.
However, the implications become troubling when a government asserts its right to appropriate a private company’s technology. This was evident when the Trump administration hinted at nationalizing tech it deemed necessary or labeling said tech as a supply-chain security risk, as Secretary of War Pete Hegseth did in his disapproval of Anthropic CEO Dario Amodei’s approach. The contradiction of labeling Claude as too dangerous, yet employing it for operations such as the capture of former Venezuelan President Nicolás Maduro and in the conflict with Iran, has not escaped attention.
Consequently, many of my students—who aspire to launch startups or develop new technologies—are left wondering: If I create something that conflicts with government policies, will it be taken from me or dismantled? Amodei opposed the use of Claude for purposes he believed could violate civil rights, such as mass surveillance or autonomous weaponry. My students contemplate how to prevent misappropriation of their creations, whether through design features, built-in safeguards, or regulatory measures. They often fear the possibility that their own government could employ their technology in ways that contravene human rights. Are they, as innovators, complicit in such potential misuse? This ethical dilemma mirrors the concerns that both Einstein and Oppenheimer faced with nuclear technology.
Stanford’s Human-Centered AI (HAI) initiative wisely advocates for an Ethics & Society Review (ESR) for those seeking project funding at scale. This review extends beyond the Institutional Review Board (IRB), which focuses only on individual subjects, to assess the broader societal repercussions of one’s work and strategies to mitigate possible harms. This leads to pivotal discussions about the ethics surrounding general-purpose versus purpose-driven AI. Some, like OpenAI’s CEO Sam Altman, controversially suggest that harms associated with general technologies cannot be predicted until products are released. Nevertheless, ESRs can provide essential frameworks for evaluating responsible, transparent, trustworthy, and accountable AI, moving beyond superficial ethics checks. I now require an enhanced version of HAI’s ESR for my course projects.
Yet, one must question the utility of ESRs in light of a president’s unilateral power to define what constitutes “legal use” of AI. What happens when governmental demands for military readiness infringe upon proprietary rights or compel individuals to compromise their ethical values? These questions are not new but have become increasingly pressing, as demonstrated by the ongoing dialogue between Anthropic and the Pentagon. As Maureen Dowd highlighted in a New York Times op-ed, the Pentagon confronted Anthropic with a stark choice: “Be extorted or blacklisted.” It is no surprise that Trump bypassed Congress in declaring conflict with Iran. The current administration has embraced a rapid, unilateral decision-making style akin to Zuckerberg’s “move-fast-and-break-things” mantra, undermining the vital legal and ethical dialogues that are necessary for informed AI policy-making.
Some propose that ultimate judgment should rest with the executive branch. However, what occurs when lobbyists from the tech industry attain unprecedented influence over presidential AI decisions? It is simplistic to assume that regulation will render the U.S. disadvantaged in an international AI competition. Assertions like “regulation stifles innovation” are often driven by motives that prioritize profit over societal good, benefiting companies eager to launch their products.
In this climate, the crucial conversations that should take place within a democracy about the ethical considerations of AI are framed as overly slow and cumbersome. Yet, discussions of this significance should unfold at a human pace, rather than in an incessant sprint. Moreover, decisions about AI governance should not fall solely to the president or CEOs. Ideally, this dialogue should involve civil society, tech professionals, academics from the humanities and STEM fields, philanthropic organizations, and all government branches—adopting a whole-of-society approach. After all, we all have a stake in the development of such transformative technologies.
— Michele Elam, William Robertson Coe Professor of Humanities in the English Department at Stanford University and Stanford HAI Senior Fellow
