PolicyA weekendIntermediate
Write a two-page policy memo on one AI risk
Pick one decision-maker, one AI risk and one decision they actually face. Lay out the options and make a clear recommendation in two pages.
Reading & writingView project
PolicyA weekendStarterFree
Check how the UK government discloses its algorithms
UK public bodies publish records about the algorithmic tools they use under the Algorithmic Transparency Recording Standard. Read five records and assess what is disclosed and what is missing.
Download PDFReading & writingData
Why it matters
Governments are rapidly adopting AI. Transparency only works if someone reads the records and asks hard questions.
Output: A short report scoring five records.
Steps
0 of 5 done-
Understand the standard
30 minRead the Algorithmic Transparency Recording Standard hub on GOV.UK: what a record must contain and why.
-
Pick five records
30 minChoose tools that affect people directly (benefits, policing, health, education) if you can.
-
Score each record
2 hrMake a checklist: purpose, data used, human oversight, known risks, how to complain. Score each record against it.
-
Spot the patterns
1 hrWhat is usually missing? Which record is the best example and why?
-
Recommend improvements
1 hrWrite three practical recommendations for the standard or the agencies.
All steps done
Share what you made, then try another project. The PDF version has room for notes.
Next: Write a two-page policy memo on one AI risk Next steps
Try next
Related projects
PolicyA few weeksIntermediate
Track AI transparency laws across countries
Which countries require AI developers to disclose safety testing, incidents or training details? Build a comparative tracker of transparency and disclosure rules.
DataReading & writingView project
Policy1–3 hoursGuided
Map the AI rules where you live
Find out which laws, strategies and guidance apply to AI in your country, and fit them onto one page.
Reading & writingDataView project
GovernanceA few weeksAdvanced
Re-check “catch-up algorithmic progress”
MIRI TGT argued that catching up to a given capability level gets cheaper by roughly 60× per year. Re-run the analysis with newer models and test how robust that number is.
DataView project
Further reading
- Algorithm Auditing Policies Rest on Flawed Assumptions About Public Sector Systems (opens in a new tab)ReadingEscher, Banovic, Green · FAccT 2026
- Foundation Model Transparency Reports (opens in a new tab)ReadingBommasani et al.
- The Alignment Project (UK AISI) (opens in a new tab)Idea listResearch areas funded by the UK AI Security Institute.
- Frontier AI Regulation: Managing Emerging Risks to Public Safety (opens in a new tab)ReadingAnderljung et al. · 2023
- Government by Algorithm: AI in Federal Administrative Agencies (opens in a new tab)ReadingEngstrom, Ho, Sharkey, Cuéllar · 2020
Related mentored projects
- [Jurisdiction-specific] Public-Sector AI Resilience Agenda (opens in a new tab)SPARChris Schmitz · Centre for Digital Governance, Hertie School
- Auditing Frontier AI Compliance via an EU Code of Practice Tracker (opens in a new tab)SPARMarkov Grey, Charbel-Raphael Segerie · CeSIA (French Center for AI Safety) / CeSIA - Centre pour la Sécurité de l'IA (French Center for AI Safety)
- Global AI Risk Observatory (opens in a new tab)SPARBart Jaworski · MATS