OpenAI, Anthropic, Google and Microsoft have all backed a proposal that would let website publishers use a single shared token in their settings to tell multiple AI companies not to use their content to train AI models.
Industry proposal · AI training opt-out
One shared token could tell multiple AI companies not to train on your content
OpenAI, Anthropic, Google and Microsoft have all backed a proposal that lets website publishers signal, through a single token in their settings, that their content should be excluded from AI model training.
Separate requests vs. one shared token
Each block is one opt-out request a publisher would otherwise manage.
4 separate requests
One per AI company
1 shared token
Read by AI developers from site settings
How the proposed signal would work
Step 1
Publisher places the token in its website settings
→
Step 2
AI developers read the token
→
Step 3
Content is signaled as excluded from model training
What is reported
Four major AI developers back the proposal
The token is a single, common mechanism
It could lower compliance costs for publishers and AI companies
Not yet confirmed
Token format and how it is verified
When it would take effect
Whether it is legally binding or only an industry commitment
Treatment of content already collected
How it interacts with existing crawler rules
Bottom line
A notable step toward standardized control over AI training data, with practical effects that depend on details yet to be published.
Watch for: technical specifications · adoption by additional AI companies · regulatory response · whether publishers see the signal as sufficient protection
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article...