k
keynews.ai
keynews.ai
WorkThe story, in brief

MosaicLeaks: Can your research agent keep a secret?

Your AI research agent just leaked your proprietary data. Here's what ServiceNow found.

Reporting by Hugging Face Blog 1 min brief
Illustration of a transparent lens revealing connected networks across layers of paper.
Exploring the next frontier of AI research.AI illustration by KeyNews
The KeyNews take

Why it matters

ServiceNow's MosaicLeaks study reveals critical security vulnerabilities in research agents—a governance and safety issue that could reshape how enterprises deploy agentic AI systems.

The key facts

5 to know

  1. Research agents vulnerable to unintended data disclosure

  2. ServiceNow conducted security evaluation of agent behavior

  3. Published via HuggingFace blog (credible venue)

  4. June 2026 publication date (recent)

  5. Implications for enterprise AI governance and safety practices

Go to the source

Hugging Face Bloghuggingface.co

Read original report
Back to today's editionMore work news

Keep reading

Related stories

More from Work
Work

Trump announces ‘AI Force’ as alarm grows over technology’s advance

Trump creates 'AI Force' — the first executive-level AI governance structure, but resists regulation.

Financial Times TechnologyWork

Nearly one in five AI researchers already expected an extinction scenario from AI back in 2024

18%. That's the average extinction-risk estimate among 1,500+ leading AI researchers in 2024 — and the number keeps climbing.

The DecoderWork

Your AI agent may have made the decision, but your company owns the risk

40% of enterprises will deploy agents by 2026. But 40% will also decommission them when governance fails. Here's what your company doesn't own.

CIO