FFTAC / SURVIVEResilience & Prevention Program

PREVENT / AI ACCOUNTABILITY

Who chose the system—and who can correct it?

The Machine is not inherently the enemy. Unanswerable decisions and poorly governed systems are the problem.

Seven public-interest questions

  1. Purpose: What does the system actually decide, rank, recommend or flag?
  2. Authority: What agency or organization is responsible?
  3. Evidence: What data and testing support deployment? What is missing?
  4. Impact: Who may be harmed by errors, bias or exclusion?
  5. Oversight: Can people detect, override or correct errors?
  6. Appeal: How can affected people receive notice, reasons, review and repair?
  7. Exit: What contingency exists if the system or vendor fails?

A small, legal civic project

Pick one publicly documented AI-related procurement by your local authority. Read contracts, public notices and impact assessments. Ask precise questions through lawful channels. Publish sources with dates, uncertainties and the agency's response. Do not claim a system is unlawful or discriminatory without sufficient evidence.

Framework alignment

NIST's AI Risk Management Framework organizes AI risk practices around Govern, Map, Measure and Manage. FFTAC's parallel language is direct: make the Machine explainable, keep the Throne answerable, protect affected people, and preserve the source.