WASHINGTON — A new federal task force on artificial intelligence was scheduled to hold its first full committee meeting Thursday, opening a 120-day push to define how the government should respond to increasingly capable systems, reported incidents and widespread public unease. The meeting matters less as a ceremonial launch than as a test of whether the administration can turn broad promises about American leadership into a clear operating plan for safety, accountability and enforcement.
The group, called the Super Intelligence Task Force by the administration, is led by Director of National Intelligence Jay Clayton. Its senior membership also includes Office of Personnel Management Director Scott Kupor, Federal Trade Commission Chairman Andrew Ferguson and Defense Department technology chief Emil Michael, according to Reuters. Kupor said the panel expected to publish a charter after its first full meeting, which should establish its goals and responsibilities. Until that document appears, the task force's practical authority, division of labor and public reporting obligations remain unsettled.
From a name change to an operating agenda
The administration has deliberately reframed the technology. President Donald Trump's Sept. 29 executive order directed executive agencies, where legally permissible, to use “Super Intelligence” or “SI” instead of “artificial intelligence” or “AI” in nonstatutory documents and communications. The order kept the existing statutory definition of artificial intelligence for now and asked the White House science adviser to propose legislative language within 60 days.
The task force has a broader assignment than terminology. Clayton has said it will deliver a report within 120 days assessing AI's risks and opportunities and recommending the federal government's role. Earlier reporting said the panel would examine how agencies receive reports of breaches, hacks and other AI-related incidents and how the government can strengthen its response using existing legal authorities.
That creates an immediate governance problem. Cybersecurity, consumer protection, intelligence, employment and defense risks fall under different laws and agencies. A useful charter therefore needs more than a list of members: it needs thresholds for when an incident becomes reportable, a process for routing information, rules for protecting sensitive data and a public method for explaining decisions. Without those elements, coordination can become a forum for discussion rather than an accountable response system.
Public concern is already measurable
The task force begins work as skepticism about AI has moved beyond specialist debate. An AP-NORC survey conducted in late September found that 64% of U.S. adults believed AI was developing too fast. Large bipartisan majorities also said keeping AI under human control and protecting workers should be important government priorities. Reuters reported separately that a recent Reuters/Ipsos poll found most voters wanted stronger government action on AI risks.
Those findings do not dictate a specific regulation, but they establish the political constraint. The public is not simply asking whether the United States can build advanced systems; many people want evidence that someone can detect failures, assign responsibility and intervene before harm spreads. The panel's early choices about transparency may be as important as its policy recommendations.
Existing tools offer a starting point
The federal government is not beginning with an empty shelf. The National Institute of Standards and Technology's voluntary AI Risk Management Framework organizes risk work around four recurring functions: govern, map, measure and manage. Its companion playbook offers practices for testing, documentation and oversight across the life cycle of a system. Those tools do not create enforcement powers, but they give the task force a common vocabulary for comparing agency and company controls.
The administration also announced a voluntary frontier-safety accord signed by major technology executives. It calls for internal monitoring of risks such as cyber and biological misuse, independent evaluation, board-level oversight and remediation of identified problems. The accord says participating companies will meet regularly to develop standards and acknowledges that some practices may eventually be codified in law or regulation.
The central question is how voluntary promises will be verified. A company can maintain strong internal controls and still face incentives to release products quickly or limit disclosure of failures. External evaluation can reduce that conflict only if evaluators have meaningful access, results are comparable and serious findings reach the appropriate regulator. The task force can add value by proposing baseline evidence that agencies and the public should expect, even when the government relies on existing authority rather than new legislation.
Incidents are rising faster than disclosure norms
The Stanford AI Index reported 362 documented AI incidents in 2025, up from 233 in 2024, while noting uneven disclosure of responsible-AI evaluations by leading model developers. Incident counts are influenced by reporting and classification practices, so they should not be read as a complete measure of harm. The trend nevertheless reinforces the need for consistent definitions and a durable reporting channel.
A credible 120-day report should answer several concrete questions: which agency receives an initial report, what severity triggers escalation, how independent experts can evaluate a system, when affected people are notified and what aggregate information will be released. It should also distinguish national-security secrecy from commercial confidentiality instead of allowing either label to block all scrutiny.
The first meeting is therefore only a procedural milestone. The task force will be judged by whether its charter establishes deadlines, owners and measurable outputs; whether it uses existing standards rather than duplicating them; and whether it gives Congress a precise account of any authority the executive branch lacks. A clear framework would not resolve every dispute over AI. It would, however, show that federal coordination can keep pace with a technology whose risks and benefits already cross traditional agency lines.
