Your business website has a job: help customers understand your services and choose your company. But should its articles, product descriptions and technical guides also help train someone else’s AI model?
That deserves a business decision, not an accidental default.
Coverage this week has highlighted a proposed shared control for expressing AI training preferences. The underlying proposal was published on October 30, 2026; it is not a finished standard or a universal switch that stops scraping. The IETF’s Training Preference Signal draft explicitly separates content-use preferences from crawling behaviour. Source: IETF Datatracker. ([datatracker.ietf.org](https://datatracker.ietf.org/doc/html/draft-kelley-aipref-training-preference-signal))
For business owners, the useful takeaway is not “block everything.” It is to decide what you want, understand the trade-offs and give someone responsibility for implementing that decision.
Three different questions
Start by separating three questions: Can an automated service retrieve your website? Can your content be used to train a generative AI model? Can it help inform an AI-generated answer?
Those are not interchangeable. Microsoft’s published guidance distinguishes training from “grounding,” which means using content to inform an answer when it is generated. Its new preferences allow businesses to make separate choices about those uses, without changing eligibility for traditional Bing search results. Source: Microsoft Bing, Your Content, Your Choice. ([blogs.bing.com](https://blogs.bing.com/search/2026/10/your-content-your-choice-clearer-preferences-for-training-and-grounding/))
Consider a consulting firm that publishes detailed educational articles. It might want those articles cited in AI answers that help prospective clients find it, while preferring that they not be used for model training. Another business might take a different approach to its public product documentation.
Neither decision should be left to whoever last installed a website plugin.
The rollout is not immediate
As of November 12, 2026, implementation is still ahead for important providers. Google says it aims to support the shared control in 2027. It also says existing company-specific controls will continue alongside it. Source: Google Search Central, Supporting a Shared AI Training Control for Websites. ([developers.google.com](https://developers.google.com/search/blog/2026/10/ai-training-control))
Microsoft says its new preferences begin taking effect as it crawls pages from February 1, 2027. Businesses currently relying on NoArchive or NoCache for training or grounding restrictions should have their web provider review Microsoft’s transition instructions before March 1, 2027. Do not assume an existing restriction will carry forward automatically. Source: Microsoft Bing. ([blogs.bing.com](https://blogs.bing.com/search/2026/10/your-content-your-choice-clearer-preferences-for-training-and-grounding/))
Ask your provider to document what works now, what is planned and when your settings need another review. Avoid treating a future commitment as a protection already in place.
Check the visibility trade-off
Before approving a blanket opt-out, involve whoever manages customer acquisition.
Google says that, unless Google-Extended is explicitly specified, it plans to interpret the shared training opt-out as disallowing Google-Extended. Its announcement also connects allowing Google-Extended with eligibility for responses and clicks through grounding in Gemini Apps. That makes provider-specific implementation important, not just the wording of the shared preference. Source: Google Search Central. ([developers.google.com](https://developers.google.com/search/blog/2026/10/ai-training-control))
Ask your website provider for a plain-English explanation of each proposed change. Which use does it restrict? Which provider honours it? Could it affect visibility in AI answers? How will the team verify that the published settings match the approved decision?
Do not accept “we turned off AI” as a complete answer.
Make a small, practical plan
Begin with an inventory of your public content. Include service pages, downloadable guides, product documentation, older campaign sites and any separate regional websites. Mark which material you actively want prospective customers to discover.
Next, write down your preferred treatment of that content. Keep the decision simple enough for both marketing and IT to understand. For example: keep public service information discoverable, review training permissions separately, and investigate anything published publicly that was intended only for customers or staff.
Then assign responsibilities. Have the business owner approve the policy, marketing review discovery implications, and the web provider or IT team implement and test the settings. Request a saved copy of the previous configuration and a record of what changed.
Finally, schedule a review as providers implement their announced controls. Ask for evidence of the live configuration rather than a verbal assurance that the work is finished.
A business decision, supported by IT
Our recommendation is to treat website AI preferences as a shared business and technology responsibility. Start with what your company wants to achieve, then choose settings that support it. Avoid both panic-driven blocking and unexamined defaults.
Magic Touch Computers is Your Global IT Department, providing business IT services since 2003 from Calgary to businesses internationally. We help bring business priorities and technology decisions into the same conversation.
To discuss your website controls and broader business IT needs, contact Magic Touch at (403) 770-6652, visit magictouchit.ca/contact, or take the free AI IT assessment at magictouchit.ca/it-assessment.
— The Magic Touch Team