Can you pass the
AI misuse quiz?
5 minutes · real findings · see if you can score 80% or better
Ten questions drawn from Anthropic’s September 2026 threat intelligence report, across all seven harm areas it covers. Pick an answer and the evidence appears immediately — why it is right, what it means for a governance programme, and the report’s own page reference so you can check it yourself.
Nothing leaves your browser. There is no account, no score submission, and no third-party quiz platform in the middle — which would be an odd thing to embed on a site that argues unassessed intermediaries are a governance problem.
Anthropic uses “uplift” to describe the capability boost an actor gains from AI. Which three dimensions does the report use to assess uplift?
Choose an answer to see the evidence
Every question and page reference comes from Anthropic, Detecting and countering misuse of AI: September 2026, published September 10, 2026 (154 pages). The quiz paraphrases for teaching and assessment; it reproduces no passage of the report. Read the original: https://www.anthropic.com/threat-intelligence-report-september-2026
The report shows the threat. The path shows the risk.
The report documents the attacks. The part drawing the least attention is a third-party risk story: labs relaying their own users’ prompts — names, corporate data, live credentials — to a model the user never chose. Our read of what that means for a mid-market governance programme, with the model-path diagram, is here.
Read: Who actually answered your prompt? →