> ## Content Index > Fetch the complete content index at: https://agarhouse.com/llms.txt > Use this file to discover other available public pages before exploring further. # Google and Bing plan a shared AI-training opt-out. Website owners still need a plan. - URL: https://agarhouse.com/google-bing-shared-ai-training-opt-out-website-owners/ - Published: 2026-11-02T13:00:00.000Z - Updated: 2026-11-02T13:00:00.000Z - Description: A shared AI-training preference could simplify website settings. Provider-specific rules and Bing’s coming transition still need a closer look. - Author: Meatball - Tags: Consumer Technology, News, Artificial Intelligence, Privacy, #agarhouse-slot-2026-11-02-0800 Can a website owner decline AI training and still have customers find the business through search? Announcements from Google and Microsoft on October 30, 2026, move toward a shared way to express that choice. They also leave owners with some homework. A shop owner or independent writer should be able to decide how their pages are used without having to become an expert in every AI company's settings. A common preference could make that job less fiddly. The coming changes deserve attention, especially if someone configured your site months ago and nobody has checked those settings since. ## A shared preference, still taking shape Google says it plans to support a shared AI-training preference in robots.txt, a site’s text file for crawler instructions, during 2027\. Microsoft has also announced support, alongside separate changes to Bing's grounding controls. These are future commitments, so owners shouldn't assume a new instruction is already being honoured everywhere. [Google's announcement](https://developers.google.com/search/blog/2026/10/ai-training-control?ref=agarhouse.com) and [Bing's implementation guide](https://blogs.bing.com/search/2026/10/your-content-your-choice-clearer-preferences-for-training-and-grounding/?ref=agarhouse.com) describe their respective plans. The proposed token, `__usage_ai-training`, appears in an October 30 Internet-Draft. It expresses a content-use preference through robots.txt, separately from crawling rules. The draft remains a work in progress, rather than an adopted final IETF standard. Leaving out the token expresses no preference through this mechanism. [The technical draft](https://datatracker.ietf.org/doc/html/draft-kelley-aipref-training-preference-signal-00?ref=agarhouse.com) sets out those boundaries. That distinction matters when deciding what to ask a developer to change. A preference needs a service willing and able to honour it. Putting text in a file doesn't create a technical barrier around your pages. ## Three different uses Start with three separate activities. Crawling means fetching pages. Training uses content to develop a generative AI model. Grounding uses material when constructing an answer. An owner might welcome someone finding a product page through an answer while declining to have that page used for training. Before touching any settings, write down what you want for each activity. “Keep us out of AI” gives a developer much less to work with than a clear preference about training, answer generation and ordinary search discovery. Google's planned handling shows why existing settings still matter: without an explicit setting for Google-Extended, its provider-specific control, it intends to treat the shared training opt-out as disallowing Google-Extended. Google also describes explicit permission for Google-Extended as making pages eligible for grounding in Gemini Apps. Owners need to review that interaction rather than assuming the shared preference replaces every provider-specific decision. ## Bing’s transition dates Bing says its new preferences will begin being honoured on crawl from February 1, 2027\. Its legacy NoArchive and NoCache signals will stop counting for these purposes after March 1, 2027\. Ordinary Bing-search eligibility will remain unchanged, subject to normal indexing and crawling requirements. Temporary grounding choices open in Bing Webmaster Tools on December 15, 2026, for sites using NoArchive or NoCache that day. To retain grounding restrictions, those owners must publish page preferences and record the temporary choice before March 1\. It expires when recrawling discovers the new preference or on September 30, 2027, whichever comes first. There is no temporary training setting. If your site relies on those older signals, give this transition a named owner and a place on the calendar. Publishing a change doesn't guarantee that Bing will discover it by a particular date. Leaving the work until the final week creates an avoidable dependency on a crawler's next visit. ## What should a small site do now? Make an inventory before making edits: your robots.txt file, relevant page tags and response headers, and any settings managed by a hosting service or plugin. Save the current configuration and record why each restriction exists. Otherwise, a well-intended cleanup can undo a choice someone made deliberately. Then compare your desired uses with each provider's current guidance. Keep the shared preference and the provider-specific exceptions in the same maintenance note, with dates for checking deployment. A short record of these decisions will also help the next person who maintains the site. The useful promise here is simpler administration. How useful it becomes will depend on implementation and clear documentation. For website owners, the immediate job is to make deliberate choices, preserve them through the transition and avoid mistaking an announcement for a finished rollout. ## Sources [Google Search Central: shared AI-training control plans](https://developers.google.com/search/blog/2026/10/ai-training-control?ref=agarhouse.com) · [Bing: training and grounding preferences and migration dates](https://blogs.bing.com/search/2026/10/your-content-your-choice-clearer-preferences-for-training-and-grounding/?ref=agarhouse.com) · [IETF Internet-Draft: proposed AI-training preference signal](https://datatracker.ietf.org/doc/html/draft-kelley-aipref-training-preference-signal-00?ref=agarhouse.com)