CYBRARY Who Was in the Training Set?
A privacy complaint arrives. Establish whether the model actually memorised the record.
Advanced · Several moving parts — expect trade-offs and a wrong turn or two.
Outcome · Investigate membership inference and privacy exposure.
AI connection
This activity runs entirely on the deterministic runtime. Optional live-model runtimes only affect dialog, coaching and wording — never checks or scoring.
Nothing leaves the browser session.
Prompts never leave the learner's machine.
Lab prompts reach the learner's chosen provider only.
Customer tenancy; Cybrary sees metadata only.
Cybrary AI gateway; retention off by default.
Per-job — the router never sends restricted jobs off-tenant.
Lab manifest
The machine-readable release contract. Authors edit this; the console refuses to publish without checks, both regression suites and a reference solve.
schemaVersion: cybrary.ai-lab/v1
id: who-was-in-the-training-set
release: 1.0.0
mode: ai-lab
maturity: 2
roles:
- privacy-engineer
- data-scientist
outcomes:
niceTasks:
- T0036
atlasTechniques:
- AML.T0024
owaspRisks:
- LLM02
aiAssets:
models:
- support-agent-model@2026.06
prompts:
- who-was-in-the-training-set-system-prompt@1
agents:
- support-agent-orchestrator@3
tools:
- crm.lookup
- ticket.update
- email.send
- refund.issue
corpora:
- support-kb@v4
enterpriseTopology: support-enterprise-v3
scenario:
seed: who-was-in-the-training-set-seed-1
injects:
- who-was-in-the-training-set-inject-1
telemetry:
requiredSignals:
- ai.trace
- rag.retrieval
- tool.call
- identity.auth
- dlp.egress
- siem.alert
evaluation:
deterministicChecks:
- who-was-in-the-training-set-checks
repeatedTrials: 20
maliciousSuite: who-was-in-the-training-set-attack-v1
benignSuite: support-utility-v4
safety:
networkEgress: deny
maxTokens: 250000
maxCostUsd: 20