OPM tells federal agencies to expand AI use in hiring
The Office of Personnel Management is urging federal agencies to use artificial intelligence more broadly in hiring while requiring documented safeguards when AI could affect who advances or receives a federal job.
OPM Director Scott Kupor issued the memorandum on Thursday, August 27, 2026, to the heads of federal departments and agencies. It says agencies should adopt AI across the hiring lifecycle rather than wait for future features in USAJOBS, USA Staffing, USA Hire, USA Class and USA Performance.
The memorandum is guidance for internal Executive Branch management. It is not a statute or regulation, and OPM says it does not create a new enforceable right or benefit for applicants.
Where OPM says AI can help
OPM identifies several uses that may generally fall outside the government’s high-impact AI category when authorized officials independently review the underlying records and remain responsible for the decision.
Examples include drafting position descriptions and job announcements, preparing evaluation statements and interview questions, supporting résumé and minimum-qualification reviews, conducting administrative quality control before a final offer, and analyzing aggregate hiring information such as applicant flow, assessment performance and time-to-hire.
An AI tool could, for example, locate and summarize information in an application. OPM says that use generally would not be high-impact when an authorized official reviews the full application, checks the source documents and makes the eligibility decision from the record rather than simply accepting the AI summary.
Human review is not automatically enough
The memorandum draws a functional line: the key question is whether the AI output becomes the principal basis for an employment decision.
A formal human approval step alone does not remove a use from the high-impact category. The agency’s record must show that an official independently reviewed the evidence, evaluated the decision logic and supplied the justification. If an official merely ratifies an AI-generated recommendation, OPM says the AI output may still be the principal basis for the decision.
AI-assisted pre-employment screening, hiring recommendations, candidate matching, scoring and ranking are presumed high-impact when the output may materially affect access to federal employment. OPM’s memorandum also points to related employment decisions involving reasonable accommodation, pay, promotion, performance management, discipline and reassignment.
What agencies must do for high-impact uses
When a use remains high-impact, the agency must implement the minimum risk-management practices in the Office of Management and Budget’s Memorandum M-25-21 or obtain a written waiver through the agency’s chief artificial intelligence officer.
OPM says agencies remain responsible for their own determinations and documentation. The memorandum’s examples are guidance on applying OMB’s framework, not a binding determination for every agency workflow.
Even a use classified as non-high-impact remains subject to other personnel and legal requirements. Before deployment and periodically afterward, agencies must verify that AI-assisted ratings, rankings and referrals correctly apply veterans’ preference; meet Rehabilitation Act accessibility and reasonable-accommodation requirements; use authorized tools with applicable privacy protections; preserve auditable records; and keep a reconsideration avenue available for adverse eligibility determinations.
OPM says that reconsideration requirement does not create a new appeal right. It requires agencies to ensure that AI-assisted steps do not block procedures that already exist, including correction processes and veterans’ preference redress.
Why OPM is pushing the change
OPM says the federal government currently takes an average of 87 days to hire and is working toward an 80-day target under the administration’s merit hiring plan. In its explanation of the policy, OPM says future AI features are planned across its USA Suite of hiring and workforce tools, but agencies should not wait for those features to begin adopting AI consistent with the memorandum.
Federal News Network reported that the guidance could make agencies more comfortable testing AI, while experts cautioned that résumé screening and candidate assessments will require validation, subject-matter review, training and sufficient agency capacity. Those concerns reflect implementation risks, not documented findings that the new policy has already produced widespread hiring failures.
What applicants should watch
Applicants may encounter AI-assisted tools used to organize applications, draft announcements or review eligibility records. Under OPM’s guidance, agencies are still expected to retain independent human responsibility for employment decisions and preserve the records supporting those decisions.
Applicants should read the instructions for any USA Hire assessment carefully. OPM’s applicant guidance, updated July 29, 2026, says using ChatGPT, Claude, Copilot or other artificial-intelligence tools during testing is prohibited unless the assessment instructions expressly allow it. Applicants must certify that they completed the assessment independently. OPM says failing to certify or falsely certifying can make an applicant ineligible for further consideration for the announcement, and dishonest conduct may lead to disqualification from federal employment.
The next test will be implementation: which agencies deploy AI, how they validate outputs, how they protect applicant data, whether veterans’ preference and accommodations are applied correctly, and whether a human decision-maker can explain an adverse eligibility decision from the underlying record rather than from an unexplained automated score.
Sources
Look for updates to this story
Discover more from Interactive News
Subscribe to get the latest posts sent to your email.