Hugging Face: The 2026 AI-Agent Attack

Lesson 6 of 6

6. Final Project: Design a Safer School Agent

Connect the case study to AI ethics and design an agent with enforceable safeguards.

🔴 Advanced 100 XP

Learn it

Design a fictional agent that summarises approved revision notes. Specify its goal, allowed tools, access limits, human approvals, monitoring and a stop procedure.

Your design should remain safe even if the agent ignores an instruction. Use approved notes, no real personal data and no live security testing. Submit your written design and reflection here.

Link your decisions to this case study and to AI Ethics or the AI × Cybersecurity course. Responsible AI design is about what a system can do, not only what it says.

Key terms

Threat model
A description of assets, threats and security boundaries.
Approval gate
An enforced review step before an action is permitted.
Kill switch
A control that stops execution and revokes ongoing access.
Safe fallback
A limited response when an action cannot be performed safely.

Case file

  1. 1Define: Specify a useful task and the information it may use.
  2. 2Constrain: Choose minimal tools and externally enforced access limits.
  3. 3Test: Walk through three fictional boundary violations.
  4. 4Reflect: Explain how the 2026 incident changed your design choices.

Try it

Sequence a responsible agent-design review.

  • 1Define the task and protected assets
  • 2Document monitoring and the stop procedure
  • 3Choose minimal tools and scoped permissions
  • 4Test fictional misuse cases and safe fallbacks
  • 5Add enforced boundaries and human approvals

Challenge

Write a 150–250 word safer-agent proposal. Include purpose, minimal tools, three enforced controls, human oversight, monitoring, a stop procedure, and three fictional tests. Cite the Hugging Face or OpenAI report and connect one decision to AI Ethics.

Pick whichever way suits you — every mode earns the same bonus XP.

Write at least 40 more characters to submit.

Mark your own work

Guided walkthrough — 0/2 clues revealed

  1. Clue 1 locked — reveal it only if you get stuck.
  2. Clue 2 locked — reveal it only if you get stuck.

Each clue costs 5 XP (never below 25 XP). You'd earn 50 XP right now.

Quiz time

Question 1 of 2Score 0

Which final design best applies the case study?

Sources & further reading

Connected courses