AI safety depends on government capacity. Washington is weakening it. - Federal News Network

A commentary published by Federal News Network on federalnewsnetwork.com in October 2026 argues that AI safety depends on government capacity, and that Washington is weakening the government capacity in question. The piece's opening example is the Cybersecurity and Infrastructure Security Agency, which it says has lost roughly a third of its workforce since early 2025 through firings and departures. The article does not give a headcount figure beyond that one-third proportion.

The commentary also cites Evan Hubinger, identified in the piece as Anthropic's alignment research lead, who has put a 10% likelihood on AI ending humanity within the next decade. Alongside that it notes that the White House's new AI task force has biosecurity as a focus and will produce a report over the next 120 days. The piece sets the two items against each other: a stated near-term risk on one side, and a review process measured in months on the other.

The commentary's stated conclusion is that "a study is not a safeguard." The article runs in the outlet's commentary section and is written as an argument rather than as a news report, drawing on the cited workforce figure and the quoted research claim to make its case about government capacity.