AnswersPublished21 sources
Cloudflare Disallow AI Training Lets Sites Block Model Training Without Leaving Search
Cloudflare’s September 15, 2026 Disallow AI Training setting separates AI training consent from search visibility: qualifying Accountable crawlers can index a site for search, while training only and non Accountable t... Apple, Google, and Microsoft were the first operators designated Accountable.
Published byEdited with GPT-5.6 TerraImages generated with GPT Image 2
Research answer
Key takeaways
- Cloudflare’s September 15, 2026 Disallow AI Training setting separates AI training consent from search visibility: qualifying Accountable crawlers can index a site for search, while training only and non Accountable t...
- Apple, Google, and Microsoft were the first operators designated Accountable. The designation is based on four current or time bound commitments around training opt outs, AI summary opt outs, URL level transparency, a...
- New Cloudflare domains have Training and Agent traffic blocked by default on pages that display ads; site owners can independently manage Search, Training, and Agent policies.
AI Prompt
openai.comCreate a landscape editorial hero image for this Studio Global article: What did Cloudflare’s September 15, 2026 launch of “Disallow AI Training” introduce for website owners, how does it allow them to prevent AI. Article summary: Cloudflare’s September 15 launch introduced **Disallow AI Training**: a setting intended to let a site refuse use of its content for model training without sacrificing ordinary search indexing. It combines an opt-out sig. Topic tags: general, general web. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts with fake numbers, clic
Cloudflare’s Disallow AI Training setting is designed to end a difficult publisher trade-off: refusing AI model training without also disappearing from conventional search results. Announced September 15, 2026, it lets a site express a no-training preference and blocks relevant crawler traffic, while preserving search access for qualifying mixed-use crawlers. 7
20
The central change: training is no longer the same as search
Cloudflare replaced its broad Block AI Bots approach with separate controls for three crawler uses:
- Search — crawling to index content for search
- Training — using content to train or fine-tune models
- Agent — traffic from AI agents acting on behalf of users
That separation allows a publisher to make different choices for each use case rather than accepting an all-or-nothing bot block. Cloudflare says the controls and recommended settings became available to publishers and businesses with the September 15 launch. 13
20
For sites that want search discovery but do not want their work used to train models, Disallow AI Training is the relevant setting. It publishes the site’s training preference through Cloudflare’s bot-policy and robots.txt tooling while applying blocking rules to training traffic. 1
6
7
How Disallow AI Training preserves search access
Some crawlers serve more than one purpose. Googlebot, Applebot, and Bingbot can crawl for search, while their operators may also have AI-related content uses. Cloudflare’s policy distinguishes that search function from training.
When Disallow AI Training is enabled:
- A qualifying Accountable mixed-use crawler can continue to access content for search indexing.
- The crawler operator is expected to honor the site’s refusal to use that content for AI training.
- Dedicated training crawlers, and mixed-use crawlers whose operators are not Accountable, are blocked under the training-disallow policy.
7
17
24
This differs from Cloudflare’s stronger Block choice. A full block can stop Googlebot, Applebot, and Bingbot, including their search crawling, so it is not the setting to use when maintaining search visibility is the priority. 17
28
What “Accountable” means
Accountable is Cloudflare’s designation for operators of mixed-use crawlers that meet, or have committed on a defined timetable to meet, Cloudflare’s standards for separating search crawling from AI-related uses. Apple, Google, and Microsoft were the first operators given the designation. 7
20
23
The designation does not mean a site has granted blanket AI permission. Instead, it is Cloudflare’s basis for allowing a mixed-use crawler to retain search access after a publisher selects Disallow AI Training.
The four Accountable requirements
Cloudflare said an operator must meet or commit to meet four conditions. The available reporting describes them as:
- A training opt-out mechanism through
robots.txtor a comparable standard. - An opt-out for AI-generated search summaries, available now or committed on a specified timeline.
- URL-level visibility into how content is used for search and training.
- A public confirmation that opting out of training will not remove ordinary search access.
23
The requirements matter because they make the search-versus-training distinction auditable in principle: site owners need both a way to state a preference and a way to understand whether it is being respected.
Which bots remain available—and which are blocked
Under Disallow AI Training, the decisive factor is not simply whether a bot belongs to an AI company. It is whether the bot is a mixed-use crawler whose operator has earned Cloudflare’s Accountable designation.
| Crawler category | Treatment under Disallow AI Training |
|---|---|
| Accountable mixed-use search crawlers, including Googlebot, Applebot, and Bingbot | May remain allowed for search indexing |
| Dedicated AI-training crawlers | Blocked |
| Mixed-use crawlers without Accountable status | Blocked for training access |
Cloudflare’s model therefore gives publishers a more precise policy than a universal AI-bot block: preserve a crawler’s search role only where its operator has committed to honor the separate training restriction.
How the controls relate to Google, Apple, and Bing
Cloudflare’s Bot Preference Sync feature keeps a site’s robots.txt aligned with AI-bot preferences set in the Cloudflare dashboard for Search, Agent, and Training. 1
6
For Google and Apple, the relevant training controls are distinct from the search crawlers:
- Google-Extended is Google’s training-related opt-out token. It is separate from Googlebot, so a training restriction can be expressed without inherently blocking Google Search indexing.
27
- Applebot-Extended is Apple’s corresponding training-related control, allowing Applebot’s search role to remain separate from an AI-training opt-out.
- Bingbot can remain available for search under Cloudflare’s Accountable model. Cloudflare’s Accountable framework treats some capabilities as commitments to be delivered on a specified timeline, rather than implying every capability is already available from every operator.
Defaults for new ad-supported domains
Cloudflare had previously announced new defaults for September 15, 2026. For new domains onboarding to Cloudflare, Training and Agent categories are blocked by default on pages that display ads. 13
That default is narrower than a whole-site search block and reflects Cloudflare’s effort to give publishers differentiated controls. The supplied sources do not establish a precise automatic migration rule for every existing customer, so site owners should review their zone-level Search, Training, and Agent settings rather than assume a particular legacy configuration changed automatically.
What this means for publishers
The practical decision is straightforward:
- Choose Disallow AI Training when you want to remain discoverable in search while denying model-training use.
- Choose a broader Block policy only when you are prepared to restrict the affected crawler’s search access as well.
- Review Search, Training, and Agent settings independently, especially for ad-supported pages and any domain with existing bot rules.
20
Cloudflare’s controls do not eliminate the need for crawler operators to honor published preferences. But the launch creates a clearer policy boundary: search indexing can remain permitted while AI training is explicitly refused, provided the crawler operator meets Cloudflare’s Accountable conditions. 7
24
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
People also ask
What is the short answer to "Cloudflare Disallow AI Training Lets Sites Block Model Training Without Leaving Search"?
Cloudflare’s September 15, 2026 Disallow AI Training setting separates AI training consent from search visibility: qualifying Accountable crawlers can index a site for search, while training only and non Accountable t...
What are the key points to validate first?
Cloudflare’s September 15, 2026 Disallow AI Training setting separates AI training consent from search visibility: qualifying Accountable crawlers can index a site for search, while training only and non Accountable t... Apple, Google, and Microsoft were the first operators designated Accountable. The designation is based on four current or time bound commitments around training opt outs, AI summary opt outs, URL level transparency, a...
What should I do next in practice?
New Cloudflare domains have Training and Agent traffic blocked by default on pages that display ads; site owners can independently manage Search, Training, and Agent policies.
Sources