US Leaders Seek Public Trust in AI Amid Global Race

Leaders from major AI companies, including Anthropic, Google, Meta, OpenAI, Nvidia, and xAI, recently met with President Trump at the White House to sign a voluntary accord on artificial intelligence. The agreement outlines commitments to stronger internal controls, independent outside evaluations, and board-level review.

Despite these efforts, a key challenge remains: earning the confidence of the American public, who have shown increasing resistance to the proliferation of AI. A recent EqualAI survey conducted by YouGov found that 46% of Americans would be uncomfortable with companies using their data with AI, regardless of safeguards. This public apprehension is seen as a national liability that could hinder America's ability to lead in the global AI race.

However, the same survey indicates potential pathways to building trust. Seventy-one percent of Americans reported that a positive experience with an AI system would increase their confidence, and 63% felt the same about positive evaluations by independent scientists. This suggests that verification, rather than mere assurances, is crucial for public acceptance.

The successful adoption of AI is vital for the nation's economy, national security, and global competitiveness. If professionals, such as nurses or manufacturers, do not trust AI-enabled tools, they will be hesitant to deploy them, limiting their widespread use and benefits.

Leaders are cautioned against calls for a complete pause on AI development, especially if such pauses do not include competitors like Chinese companies. This could allow technology to advance elsewhere without the desired guardrails. The critical question facing the U.S. is how to foster sufficient public confidence to encourage AI adoption.

The public's sentiment is reflected in a Reuters/Ipsos poll, where 73% of Americans believe AI companies have not done enough to prevent harm, and 55% support slowing AI development. The EqualAI polling further highlights the desire for external validation, with 63% wanting independent scientific review and 60% supporting third-party audits. Additionally, a strong majority believe it is essential to be able to correct AI errors and appeal AI-driven decisions.

During testimony before the House Select Committee, a clear bipartisan interest in establishing robust AI governance was evident. The message conveyed was that American leadership in AI requires leadership in governance. Congress has existing authorities, such as procurement, agency governance, and consumer protection, that can be leveraged. Clarifying AI-specific protections, defining protocols for AI incident reporting akin to cybersecurity, establishing accountability for high-risk AI use cases, and prioritizing AI literacy are seen as essential steps.

Companies can also take proactive measures. Four key steps include making AI deployment visible within organizations to understand its use and impact, establishing independent evaluation processes with clear benchmarks and failure protocols, creating systems for identifying and handling significant AI incidents with defined escalation pathways, and requiring accountability by assigning point persons for certain risk categories. As AI becomes more agentic, ensuring systems have only minimum necessary access is paramount.

Finally, fostering AI literacy among the workforce and the general public is crucial. Individuals do not need to become engineers but should understand AI's capabilities and limitations, know when to question it, and recognize when human judgment must prevail.

Historically, technological advancements like automobiles did not halt progress due to the introduction of safety standards, licenses, and traffic laws. Instead, these institutions enabled mass adoption. AI is expected to follow a similar trajectory, where effective governance accelerates trust and, consequently, adoption.

The path forward involves not halting progress to eliminate every risk, nor asking Americans to blindly trust AI. Instead, the development of AI capabilities must be matched by the development of governance mechanisms that enable public confidence and widespread adoption. America's historical advantage lies not just in creating powerful technologies, but in building the institutions and standards that allow them to be trusted and adopted at scale.