Provision record
Google DeepMind · Google DeepMind Frontier Safety Framework · View original document ↗

Government Disclosure of Unmitigated Material Safety Risk

High severity High confidence Explicit document language Common · 272 of 352 platforms

Key Facts

When does Google DeepMind aim to share information with appropriate government authorities?
Google DeepMind aims to share information with appropriate government authorities if it assesses that a model has reached a critical capability level that poses an unmitigated and material risk to overall public safety.
What does Google DeepMind assess poses an unmitigated and material risk to overall public safety?
Google DeepMind aims to share information with appropriate government authorities if it assesses that a model has reached a critical capability level that poses an unmitigated and material risk to overall public safety.
Stay ahead of the changes
Track Google DeepMind and get the diff the day its terms change.
Share 𝕏 Share in Share 🔒 PDF

This analysis describes what Google DeepMind's agreement states, permits, or reserves. It does not constitute a legal determination about enforceability. Regulatory applicability and practical outcomes may vary by jurisdiction, enforcement context, and individual circumstances. Read our methodology

ConductAtlas Analysis

Why it matters (compliance & governance perspective)

The obligation is qualified by 'aim to' rather than an unconditional commitment, and is triggered only by Google DeepMind's own assessment that both 'unmitigated' and 'material' risk thresholds are met, which affects the reliability and scope of the disclosure commitment.

Interpretive note: The excerpt ends with an ellipsis, suggesting the full provision may include additional conditions, recipients, or qualifications not available for analysis. The phrase 'aim to' is a softer commitment than 'will' or 'shall' and has been preserved exactly.

Consumer impact (what this means for users)

Readers are informed that government disclosure is not automatic or unconditional—it is aspirational ('aim to') and depends on Google DeepMind's own assessment that a specific dual-threshold of unmitigated and material public safety risk has been reached.

How other platforms handle this

Tinder Medium

Consistent with applicable law, you understand and agree that Tinder may not share information with you regarding your account termination if doing so would potentially impair the safety or privacy of other users.

Leonardo AI Medium

We will make it clear by notice to you which (if any) goods or services, or website links, we receive a benefit from by featuring them on our Platform.

Glassdoor Medium

In certain situations, Glassdoor may be required to disclose personal data in response to lawful requests by public authorities, including to meet national security or law enforcement requirements.

See all platforms with this clause type →
▸ View Original Clause Language DOCUMENT RECORD
"
if we assess that a model has reached a CCL that poses an unmitigated and material risk to overall public safety, we aim to share information with appropriate government authorities...

Excerpt from Google DeepMind's Frontier Safety Framework

Provision details

Document information
Document
Google DeepMind Frontier Safety Framework
Entity
Google DeepMind
Document last updated
July 5, 2026
Tracking information
First tracked
July 6, 2026
Last verified
July 9, 2026
Record ID
CA-P-068356
Document ID
CA-D-00919
Evidence Provenance
Source URL
Wayback Machine
Content hash (SHA-256)
0484a7e766b88c19c64336c5111da9d27b2ccf083df3ba55340b35f7dcb6842d
Analysis generated
July 6, 2026 15:48 UTC
Methodology
Evidence
✓ Snapshot stored   ✓ Hash verified
Citation Record
Entity: Google DeepMind
Document: Google DeepMind Frontier Safety Framework
Record ID: CA-P-068356
Captured: 2026-07-06 15:48:00 UTC
SHA-256: 0484a7e766b88c19…
URL: https://conductatlas.com/platform/google-deepmind/google-deepmind-frontier-safety-framework/provision/CA-P-068356/government-disclosure-of-unmitigated-material-safety-risk/
Accessed: Aug. 19, 2026
Permanent archival reference. Stable identifier suitable for legal filings, compliance documentation, and research citation.
Classification
Severity
High
Categories

Other risks in this policy

Get the research letter

Companies change their terms quietly. We read every version and catch what actually changed. One email a week on the changes that matter and what they mean.

Frequently Asked Questions

What does Google DeepMind's Government Disclosure of Unmitigated Material Safety Risk clause do?

The obligation is qualified by 'aim to' rather than an unconditional commitment, and is triggered only by Google DeepMind's own assessment that both 'unmitigated' and 'material' risk thresholds are met, which affects the reliability and scope of the disclosure commitment.

How does this clause affect you?

Readers are informed that government disclosure is not automatic or unconditional—it is aspirational ('aim to') and depends on Google DeepMind's own assessment that a specific dual-threshold of unmitigated and material public safety risk has been reached.

How many platforms have this type of clause?

ConductAtlas has identified this type of provision across 272 platforms. See the full comparison.

Is ConductAtlas affiliated with Google DeepMind?

No. ConductAtlas is an independent monitoring service. We are not affiliated with, endorsed by, or sponsored by Google DeepMind.