White House Announces AI Responsibility Accord, Executive Orders, and Super Intelligence Task Force

In June, the president, seemingly in reaction to the cybersecurity risks of frontier artificial intelligence models, ordered the establishment of a voluntary regulatory regime for the government to scrutinize frontier models prior to public release.

In the past week, amid still rising concerns over the safety of AI models and mounting calls to regulate against their dangers, the White House apparently went in the other direction, endorsing a light regulatory approach that the president characterized as corporate “self-policing.”

On September 29, President Trump signed a “White House Accord on Super Intelligence” (Accord) with half-a-dozen tech industry leaders. They voluntarily agreed to adopt safety policies that include audits and third-party evaluations and risk reviews. 

The same day, the president issued two executive orders. “Inaugurating the Era of Super Intelligence” directed the Executive Branch, “to the maximum extent permitted by law,” to adopt the term “super intelligence” instead of “artificial intelligence” in correspondence, communications, and “non-statutory documents.” The White House intends the move to “capture[] the promise, potential, and rapidly advancing capabilities of these technologies.”

The executive order “Streamlining Access to Government Services Through America.gov” revamped the America.gov website to streamline Americans’ access to information and services speckled across government websites through an AI chatbot.

And, on October 4, the president announced the creation of an interagency “Super Intelligence Force” (SIF) to “coordinat[e] the effort of the Federal Government to ensure that America continues to lead the World in Super Intelligence.” The SIF will coordinate with consumers, public interest groups, religious organizations, critical infrastructure providers, and AI companies and issue a report in 120 days on the risks and opportunities presented by AI, according to reports.

White House Accord on Super Intelligence

After meeting in the White House, executives at six leading AI companies signed the one-page Accord, agreeing voluntarily to adopt safety policies that include four layers of controls and audits, in addition to any other precautions:

1. Internal controls to monitor models during training and deployment in areas such as cybersecurity, biosecurity, and chemical threats, and to prevent models from hacking or gaining unintended access to technical systems.
2. An internal team to ensure controls, monitoring, and detection are operating properly, and to fix any issues.
3. An independent external auditor to assess the proper operation of controls, monitoring, and detection.
4. An independent committee of the board of directors to oversee and receive reports from internal teams and independent auditors.

Implications

• Voluntary vs. Compulsory Action. The Accord does not support codifying these commitments in legislation immediately, saying only it may make sense “over time” to turn these steps into law.
  
  Public-private frameworks like the Accord have been a common approach to federal AI policy. Former President Biden brokered voluntary safety commitments among similar such corporations in July and September 2023. Three years later, AI models are much more powerful and their risks, especially recent agent-enabled cyber breaches, have heightened concern of AI safety, helping to prompt a wave of state lawmaking to require transparency plans, critical incident reporting, and, in Illinois, safety audits. While these are state laws, Congress is considering the FRONTIER Act, which includes many of the principles of the Accord and state counterparts, including robust internal risk management, strong corporate governance, and independent third-party evaluation. For their part, many AI developers already have internal safety teams, and some AI executives have been echoing the Accord’s concerns for weeks.
• Audits. The Accord’s call for partnering with an “independent external auditor” is redolent of new laws signed this month in California that create an AI auditor validation process.
• Governance. The Accord’s pledge to designate an independent committee of the board to “oversee” the internal and external auditors is consistent with a broader embrace of board responsibility of AI deployments. For example, last year, the Investor Advisory Committee recommended to the Securities and Exchange Commission that it require issuers to disclose their boards’ oversight of AI deployments.
• Reach: While signed by only six companies, the Accord asserts that “every company is responsible for developing its own technology safely and in a way that builds trust with customers and the public,” and it purports to speak for those “training and deploying frontier models.” Those “training” AI is a relatively small number. Those “deploying” models is much greater. Whether deployers will adopt this Accord remains to be seen.
• Cooperation. The Accord states that the signatories will “meet regularly to establish standards.” This move may raise antitrust concerns; plaintiffs have already filed antitrust suits over their alleged coordination to “pace” AI development.

Executive Orders

“Inaugurating the Era of Super Intelligence” mandates that the executive branch use the terms “Super Intelligence” and “SI” in place of “Artificial Intelligence” and “AI,” and the order says the Trump Administration will no longer “acknowledge the usage of ‘Artificial Intelligence’ or ‘AI’ in any applicable setting.” The order is not retroactive and does not require alteration of “previously issued regulations, Presidential actions, contracts, grants, or other historical documents.” The term SI may have limited public appeal; a recent poll showed that 53% of respondents considered “artificial intelligence” the preferred term while 9% favored “super intelligence.”

“Streamlining Access to Government Services Through America.gov” establishes a digital point of entry through which individuals in the U.S. can access federal information, services, and transactions. The order states that America.gov will provide access to “covered services,” that is, public-facing federal services that serve more than 100,000 users per year and are accessible online, excluding IRS tax filing services, Department of War services, and elements of the Intelligence Community. Each agency should preserve control of its respective records, systems, statutory responsibilities, and adjudicatory authority.

Implications

• Nomenclature. The substantive import of “Inaugurating the Era of Super Intelligence” remains to be seen, though government contractors may need to be adaptive as agency contracts and requests for information incorporate updated terminology. The order directs the preparation of legislative language to establish a federal definition of “Super Intelligence” within 60 days. Should such legislation become law, businesses will need to note its impact on the scope of the existing definition for “Artificial Intelligence” under 15 U.S.C. § 9401(3).
• Possible Opportunities. As agencies fund technology integration with America.gov, new contracting opportunities may emerge. Businesses bidding on these engagements can prepare for the zero data retention policy that the program requires by auditing existing data flows and vendor contracts. Moreover, businesses may need to assess the accuracy of their products to comport with the legally significant tasks that government services demand.

Super Intelligence Force

The SIF will be led by Director of National Intelligence Jay Clayton, Chairman of the Federal Trade Commission Andrew Ferguson, Under Secretary of War for Research and Engineering and Chief Technology Officer Emil Michael, and Director of the Office of Personnel Management Scott Kupor. The SIF will have four months to coordinate with, in the president’s phrase, “Consumers, Public Interest Groups, Religious Organizations, Critical Infrastructure Providers, and Super Intelligence Companies” and write a report on AI’s risks and opportunities.

Implications

Industry may have opportunities to engage with the SIF. While the SIF goes about its work, private plaintiffs, state attorneys general, federal regulators including the Federal Trade Commission and Securities and Exchange Commission, congressional committees, state legislators, and foreign governmental bodies are acting with increasing aggressiveness to investigate alleged harms, in light of evolving technological capabilities and darkening public perceptions. AI developers now face layered and evolving compliance obligations that are unlikely to abate in the near term.