AI in government
Governments use AI to help deliver services, process paperwork, and analyze data. Government decisions can affect people’s rights and access to benefits, so AI use must be careful and accountable. This page explains what “AI in government” means and where it shows up. This is educational only.
What “AI in government” means
In government, AI can mean prediction and classification models (like fraud flags) and generative AI (like drafting a letter). These tools often support humans, but they can also influence decisions, so oversight matters (see Executive Order 14110).
Common uses (where it shows up)
- Public service chatbots and routing: answer common questions and send people to the right department.
- Document processing: extract data from forms and help staff review applications.
- Fraud detection with oversight: flag unusual cases for human review.
- Drafting letters: draft a plain-language notice with tools like Grammarly or DeepL Write (review required).
- Summaries: summarize long text with tools like Otter or Notion AI.
- Translation and accessibility: translate content and create simpler versions for accessibility.
Dive deeper with BonsAI Chat
What AI is good at (and bad at)
- Good at: summarizing long documents and spotting patterns in data. (NIST AI RMF)
- Good at: routing requests and helping staff find information faster.
- Bad at: making fair decisions on its own. People must stay accountable.
Risks you must take seriously
- Made-up facts: the model can sound confident and be wrong. (NIST AI 600-1)
- Rights impact: a bad model can affect benefits, services, or enforcement.
- Bias: outputs can treat groups unfairly.
How to use AI safely (simple checklist)
- Do not paste private citizen data into tools you do not control.
- Use AI for drafts, then do human review before anything is sent.
- Test on edge cases and monitor errors. (OWASP LLM Top 10)
How rules and regulators think about it (high level)
- Old rules still apply (privacy, anti-discrimination, records). (OECD AI Principles)
- For high-impact uses, people expect transparency and strong controls.
Questions to ask before you trust a tool
- What data did it use, and can we audit it? (NIST AI RMF)
- Can a person appeal or override the result?
- Can we explain the decision to the public?