Item detail

MosaicLeaks: Can your research agent keep a secret?

MosaicLeaks is a new ServiceNow research release showing how deep-research agents can leak sensitive facts through seemingly ordinary external searches, plus a privacy-aware training approach aimed at reducing that behavior.

Score8.1
Popularity72.0
Risknone
TierSilver
Score breakdown
Usefulness8.0
Novelty8.0
Momentum7.0
Maturity7.4
Open-source/build6.8
Evidence7.2
Workflow potential8.9
Setup ease4.2

Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.

Why it matters

That matters because research agents are being pointed at private documents and live web tools at the same time, and privacy failure in that setting is much more concrete than the usual vague warnings about AI mistakes.

Who should use it

agent builderssecurity researchersenterprise AI teamspeople evaluating deep-research workflows

Who should skip it

Skip if the source link, docs, or setup requirements do not match your workflow.

Risk explanation

No inherent user-impacting risk is flagged from the captured evidence.

Evidence links

Closest alternatives / related signals

agent-safetyprivacyresearch-agentssecuritybenchmark