GAO warns poor government data and skill gaps could make AI produce 'bad outcomes faster'

A senator asked GAO how artificial intelligence will affect the federal workforce and whether the government is prepared to manage the technology’s risks.

Mister Dodaro told the committee that AI will have “enormous implications” across both white‑ and blue‑collar work and cautioned that poor government data will make AI produce “bad outcomes faster.” He said GAO has completed roughly 50 studies of AI since 2018 and developed a framework for auditing algorithms, but warned that agencies often lack complete, accurate and reliable data needed for safe AI deployment.

Why it matters: As federal agencies pursue AI to improve services, poor inputs or weak oversight can amplify error and fraud. GAO underscored the need for training programs, better recruiting (particularly for cybersecurity and AI specialists) and stronger implementation of existing laws requiring fraud risk assessments and data analytics.

Mister Dodaro also flagged future technical risks, saying quantum computing could eventually undermine current cryptography and that fraud actors are already exploring AI tools to game government systems. He urged Congress and agencies to accelerate education, workforce development and implementation of fraud‑risk frameworks to reduce vulnerabilities.

The hearing concluded with GAO offering recommendations rather than new regulatory proposals; lawmakers signaled interest in translating GAO’s recommendations into oversight steps and potential training and hiring initiatives.