Standards & Corrections
Govbase’s job is to show you what the government is doing. That only works if you can trust how we produce what you read. This page explains exactly that.
AI-generated content
Summaries, briefings, story updates, plain-language status sentences, and impact analyses on Govbase are generated by AI systems reading primary sources: bill text from Congress.gov, Federal Register documents, recorded votes, and reporting from rated news outlets. Generated text is constrained to cite its sources inline, use absolute dates, and avoid editorializing. It is not reviewed by a human before publication.
AI systems make mistakes. Every generated surface links to the underlying primary sources; for anything that matters to you, verify against the official record.
Corrections
When we learn that something we published is wrong, we fix it promptly and regenerate the affected summaries from source. To report an error, use the contact page and include the page URL. Reports about factual claims, vote tallies, bill statuses, and misattributed positions are triaged first.
Bias ratings on news sources
The Left / Center-Left / Center / Center-Right / Right labels on news articles rate the outlet, not the individual article. Ratings are consolidated from published assessments by AllSides, Media Bias/Fact Check, and Ad Fontes Media. Outlets those services have not rated appear without a label. Ratings describe typical framing and story selection; they are not a judgment of accuracy.
Nonpartisanship
Govbase does not take positions on policy. Generated text follows a house style that requires parallel treatment of both parties’ actions, bans loaded verbs and editorial qualifiers, and attributes every characterization to a named source. Where data cuts one way, we present the data and its source and stop. See also our methodology for how impact scores are computed.