This comes after the CMA gave the platform “Strategic Market Status” under its new digital competition regime last year, ruling that Google has substantial market power.

The publisher conduct requirement is intended to address the lack of meaningful choice for publishers; insufficient transparency over how content is collected for search and used in Google’s generative AI responses; and the lack of appropriate attribution when Search Content is used.

Specifically, the proposed publisher requirements are: 

1. Effective controls 
    
   Google will be required to provide “effective controls”, allowing publishers to withhold their Search Content from being used in the training and grounding[1] of broader generative AI services. As well as from the grounding of Google’s search generative AI features. Google must ensure these controls evolve appropriately as generative AI services and features develop. 
    

2. Transparency and metrics 
    
   Google will be required to publish clear information explaining how Search Content is used for training and grounding across its generative AI services and features, including the scope and effect of the available controls. Google must also provide publishers with clear, detailed metrics on user engagement where their Search Content is used in search generative AI features.  
    

3. Attribution and factuality 
    
   Google will be required to take reasonable steps to ensure Search Content is sufficiently attributed when used in generative AI search features, and to publish clear information explaining the measures it has taken to ensure and assess the factual accuracy of those features. 

The proposed conduct requirement also cover data portability, fairness in search rankings, and user choice architecture.   

The CMA is running a consultation on the Conduct Requirements, which closes on 25 February 2026. The PPA’s Head of Policy and Public Affairs, Eilidh Wilson, is leading on a response.  

Sajeeda Merali, PPA, CEO made the following statement: 

“Publishers are looking for proof, proper checks, and practical enforceability. The wish list is simple: clear data on how AI affects visibility, reliable guarantees that opting out won’t disadvantage them, and an auditable system so we can independently verify what’s happening in practice. 

On visibility, publishers want to know when their content is used in AI overviews, how it is attributed, and what traffic patterns follow. It’s vital that Google builds a reporting infrastructure that delivers the granular, auditable data that publishers are looking for. 

This means providing verifiable, like-for-like performance data covering both classic search and AI overviews. Publishers will need to see stable indexing, unchanged rankings, and no reduction in referral quality or volume. Anything less leaves too much uncertainty for an already fragile digital advertising environment. 

Self policing alone would not be sufficient. These measures only work if they are backed by clear obligations, external oversight, and enforceable consequences for non-compliance. That’s precisely why the CMA’s proposed conduct requirements matter: it introduces independent scrutiny for the first time and must be supported by an infrastructure that will enforce the requirements. 

Opt out is an essential safeguard, but it doesn’t resolve the wider value exchange question. AI Overviews still replace clicks in many contexts, and without a clear model for licensing, the commercial imbalance remains. Today’s proposal is a major step forward, but it is not the end of the conversation on publisher compensation. 

Overall, this consultation provides an encouraging foundation for a fairer digital marketplace. We will continue to engage constructively with the CMA and our members throughout the upcoming consultation period to ensure the final requirements deliver the clarity, safeguards, and accountability that publishers urgently need.” 

If you have any questions, or would like to find out more, please get in touch: eilidh.wilson@ppa.co.uk 

Footnote: [1] the PPA interprets grounding as the process of connecting an AI mechanism to an external data point (such as publisher Search Content) in real-time for features such as fact- checking.