Skip to content
AI Security

Interview question · Software, Data & Model Supply Chain

AI coding assistants sometimes suggest packages that do not exist. Why is that a supply-chain risk, and how do you control it?

Typosquatting waits for a human typo. Slopsquatting waits for a model to invent the same wrong name again, and models repeat themselves.

mediumNewPreparation tags: Microsoft, Google, Hugging FaceUpdated Nov 2026

Typosquatting waits for a human typo. Slopsquatting waits for a model to invent the same wrong name again, and models repeat themselves.

Make your first attempt

What would you say before reading the answer?

  1. 01 State the assumptions your answer depends on.
  2. 02 Explain the mechanism and choose a control.
  3. 03 Name a condition that would change your decision.

The worked answer includes an approach, scoring criteria and follow-up questions. Compare the reasoning with your own.

10 answers per topic instead of 5, plus saved progress and bookmarks · costs nothingor unlock all 529 remaining answers · $25

Every worked answer in all 17 topics, every lesson in every course, the full concept curriculum, the complete practice-test bank, and the downloadable security field guides.

6 months · One payment · No auto-renewal

What if · the follow-ups in this answer

The answer works through 6 follow-ups like an interviewer would

Developers at your company use an AI assistant that often writes install commands, and a new coding agent can run them itself in the project environment.

  1. Why is a hallucinated package name more dangerous than a random typo?
  2. What if the developer checks that the package exists before installing?
  3. What if the agent runs installs itself in the developer's environment?
  4. What if all installs already go through an internal mirror with an allowlist?
  5. Can you reduce hallucinations at the assistant level too?
  6. What dependency policy would you set for AI-assisted development?

Each rung has a nudge, a model answer, the principle that carries over and the weak answer to avoid. Premium opens every answer and ladder.

UP NEXT ON YOUR JOURNEY
Next in this trackA self-spreading worm is publishing malicious versions of packages you depend on. How do you respond?Next in this trackHow can untrusted issue and pull request text turn an AI agent in CI into a supply-chain attack, and how do you run one safely?Next in this trackAll release checks are green, but they assessed different revisions. Can the release proceed?
DISCUSSION · 0

No comments yet. Share a correction or a different way to answer it.