PUBLISHED ASSESSMENT
Google Joins OpenAI And Anthropic In Promising New Scraping Controls
This compact public record contains an assessment and source links. It does not contain article text, reader identity, browsing history, or private notes.
The article accurately reports the shared proposal, syntax, participants and announced timelines, but repeatedly mischaracterizes a content-use preference as a control that prevents scraping. It also overstates Anthropic's present readiness and omits the draft's lack of formal IETF standing. Attribution is clear, though inspectable source links and important implementation context are absent.
Evidence support
Mixed (58/100)
- The participating companies, syntax, Microsoft date, Google timing and OpenAI timeline are supported by primary records.
- The article materially conflates a content-use preference with a crawling prohibition. The draft explicitly says it is not a crawling directive.
- Anthropic had started implementing support, but said deployment across its services would take time; saying it already supports the signal overstates readiness.
- Google-Extended is a policy token governing use, not a distinct crawler user agent.
Balance & context
Mixed (54/100)
- The article usefully notes that the signal is not retroactive, does not remove pages from Google Search and has staggered implementation dates.
- It does not clearly explain that the proposal is an unendorsed Internet-Draft with no formal IETF standing.
- It omits the key distinction between controlling fetching and expressing preferences about subsequent use.
- It gives limited attention to voluntary compliance, provider-specific interpretations and Google's inclusion of grounding under Google-Extended.
Transparency
Mixed (52/100)
- The article provides a named author, publication date, exact syntax and attributed implementation timelines.
- It identifies the proposal's author and participating organizations.
- The supplied article contains no inspectable external source links to the draft or company announcements.
- It does not identify the draft number or disclose its preliminary standards status.
Confidence and coverage
Confidence in this analysis: High confidence
Evidence coverage: Substantial coverage
- Actual future compliance by the companies cannot yet be independently tested.
- Anthropic's implementation status is documented through the proposal author's IETF message rather than a separate corporate announcement.
- The central claims were compared with the IETF draft and contemporaneous statements from Google, OpenAI and Microsoft.
- The use-versus-crawling distinction is explicit in the proposal itself.
- An independent trade publication corroborated Microsoft's announcement and dates.
Sources considered
- Machine-Readable Training Preference Signal — Supports the syntax but states that the token expresses a usage preference, is not a crawling directive and has no formal IETF standing.
- New draft: draft-kelley-aipref-training-preference-signal-00 — The author says Anthropic has started implementing support but that deployment across services will take time.
- Supporting a shared AI training control for websites — Confirms Google's commitment, shared-token interpretation and 2027 implementation target.
- Google's common crawlers — Explains that Google-Extended is a policy token rather than a separate crawler and that it governs training and grounding without affecting Search inclusion.
- Making it easier for websites to express AI training preferences — Confirms OpenAI's voluntary commitment and intended implementation in the coming months.
Supporting source link could not be verified automatically. Access may be restricted or the page may have changed. A recheck is scheduled.
- Your content, your choice: clearer preferences for training and grounding — Confirms Microsoft's February 1, 2027 start date and documents additional page-level controls.
- Microsoft Bing gives publishers more control over how AI uses their content — Independently corroborates Microsoft's dates and distinguishes training preferences from grounding controls.
Supporting source link could not be verified automatically. Access may be restricted or the page may have changed. A recheck is scheduled.
This assesses evidence and presentation; it is not an overall truth verdict.
Reassess
⚠ Paid reassessment incurs a charge. Opening this page does not start a run or incur a charge.
To request a new assessment, sign in to My library, watch this record, and select Request new assessment. Requests await separate approval and do not themselves start a paid run.
A new assessment may differ. Earlier published results remain in history.
Open My libraryAssessment history
Published assessment trend
Available public snapshots only; this is not a count of all executions. Corrections may include earlier scores. Gaps mean not scored or not recorded. Different scores do not establish changed source evidence.
| Date | Evidence support | Balance & context | Transparency |
|---|---|---|---|
| 2026-10-31 (current) | 58 | 54 | 52 |