Provision record
Anthropic · Anthropic Privacy Policy (Superseded Capture) · View original document ↗

Safety-flagged content training override

High severity High confidence Explicit document language Common · 279 of 352 platforms

Key Facts

Will Anthropic use a user's Inputs and Outputs for model improvement when conversations are flagged for safety review, even if the user has opted out?
Even if a user has opted out of model training, Anthropic will use that user's Inputs and Outputs for model improvement when those conversations are flagged for safety review or when the user has explicitly reported the materials to Anthropic.
Stay ahead of the changes
Track Anthropic and get the diff the day its terms change.
Share 𝕏 Share in Share 🔒 PDF

This analysis describes what Anthropic'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 opt-out for model training is not absolute; two specific conditions—safety flagging and user-initiated reporting—override the opt-out, meaning users cannot prevent training use of flagged or self-reported content.

Clause Stability Stable

0
Changes
5
Months Monitored
Jul 10, 2026
First Seen
Jul 10, 2026
Last Seen
This clause type exists across 3333 other provisions on other platforms.

Consumer impact (what this means for users)

An opted-out user's Inputs and Outputs will still be used for model improvement if those conversations are flagged for safety review or if the user explicitly reported the materials to Anthropic.

How other platforms handle this

Roblox Medium

Where the law allows us to, we may use the content you and other users have posted for training or to help us to improve the way we filter content on our platform.

Notion Medium

In accordance with applicable legal requirements, we may use your information to tailor and provide you with content and advertisements.

Glassdoor Medium

We may use your personal data to personalize your experience interacting with Content, including what Content we recommend, show, or don't show to you.

See all platforms with this clause type →
▸ View Original Clause Language DOCUMENT RECORD
"
Even if you opt-out, we will use Inputs and Outputs for model improvement when: (i) your conversations are flagged for safety review...or (ii) you've explicitly reported the materials to us

Excerpt from Anthropic's Privacy Policy (Superseded Capture)

Applicable regulations

EU AI Act
European Union
BIPA
Illinois, USA
CCPA/CPRA
California, USA
Colorado AI Act
US-CO
CAN-SPAM
United States Federal
ePrivacy Directive
European Union
EU AI Act - High Risk Provisions
EU
FTC Act Section 5
United States Federal
GDPR
European Union
UK GDPR
United Kingdom

Provision details

Document information
Document
Anthropic Privacy Policy (Superseded Capture)
Entity
Anthropic
Document last updated
May 5, 2026
Tracking information
First tracked
July 9, 2026
Last verified
July 9, 2026
Record ID
CA-P-017425
Document ID
CA-D-00012
Evidence Provenance
Source URL
Wayback Machine
Content hash (SHA-256)
e91b78d120f18b8a635385fb036a9ad6b0135fe530a2e4aadcc4d575da32fca0
Analysis generated
July 9, 2026 17:12 UTC
Methodology
Evidence
✓ Snapshot stored   ✓ Hash verified
Citation Record
Entity: Anthropic
Document: Anthropic Privacy Policy (Superseded Capture)
Record ID: CA-P-017425
Captured: 2026-07-09 17:12:50 UTC
SHA-256: e91b78d120f18b8a…
URL: https://conductatlas.com/platform/anthropic/anthropic-privacy-policy-superseded-capture/provision/CA-P-017425/safety-flagged-content-training-override/
Accessed: Aug. 18, 2026
Permanent archival reference. Stable identifier suitable for legal filings, compliance documentation, and research citation.
Classification
Severity
High
Categories

Other risks in this policy

Related Analysis

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 Anthropic's Safety-flagged content training override clause do?

The opt-out for model training is not absolute; two specific conditions—safety flagging and user-initiated reporting—override the opt-out, meaning users cannot prevent training use of flagged or self-reported content.

How does this clause affect you?

An opted-out user's Inputs and Outputs will still be used for model improvement if those conversations are flagged for safety review or if the user explicitly reported the materials to Anthropic.

How many platforms have this type of clause?

ConductAtlas has identified this type of provision across 279 platforms. See the full comparison.

Is ConductAtlas affiliated with Anthropic?

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