Search Engine Roundtable reports that Google announced plans to voluntarily adopt an Internet Engineering Task Force proposal that will create a streamlined AI training control for creators and web publishers. Scheduled for 2027, the initiative lets websites express preferences across multiple AI providers in a single step.
Key takeaways
- Google joins Bing, OpenAI, and Anthropic in adopting the proposed IETF standard.
- The shared policy token in robots.txt will manage training preferences in one step.
- Google intends to support the signal using user-agent: __usage_ai-training in early 2027.
- Existing company controls like Google-Extended will continue to function alongside the new signal.
How will the new AI training control work?
Search Engine Roundtable explains that website owners can use a shared policy token within their robots.txt file instead of listing each developer separately. Google stated that this will help websites express their AI training preferences across multiple AI providers in a single step and is based on work from the AIPREF working group at the Internet Engineering Task Force.
When a site opts out using this policy token and does not explicitly specify Google-Extended, Google plans to interpret that choice as a full opt-out. Other AI providers can handle the signal in a similar fashion. John Mueller noted that existing company-specific instructions will stay active alongside the joint signal.
Managing robots txt and Google Extended
Websites that specifically allow Google-Extended remain eligible to appear in responses and receive clicks via grounding in Gemini Apps. Search Engine Roundtable mentions that Google also maintains the Search Console opt-out method specifically for Search generative AI. This draft proposal aligns with the EU Code of Practice for General Purpose AI Model Providers under the EU AI Act.
Why this matters for web publishers
Publishers, civil society groups, and regulators have called for easier industry-wide options to express machine-readable preferences. This move gives website owners early and expanding access to a cross-industry standard while discussions on the rest of the AIPREF vocabulary continue. Google aims to roll out support alongside other providers in 2027, providing an easier path to manage permissions without updating separate instructions for every single developer.
Frequently asked questions
Google plans to implement and support the new AIPREF shared machine-readable training preference signal in early 2027.
Google added a note indicating support for the signal using the user-agent: __usage_ai-training.
Yes, existing company-specific controls like Google-Extended will continue to function alongside the new joint industry signal.
Google, along with Bing, OpenAI, and Anthropic, plans to adopt the proposal under consideration at the Internet Engineering Task Force.
Websites that specifically allow Google-Extended remain eligible to appear in responses and receive clicks via grounding in Gemini Apps.
Based on reporting by Search Engine Roundtable. Read the original article
