Brand and Creative
Content provenance and AI disclosure
This is the narrowest useful unit of Brand and Creative: keep AI-assisted work honest now that the platforms check for it.
Content provenance and AI disclosure, defined
Keep AI-assisted work honest now that the platforms check for it. In practice that means watermarking for AI-generated visuals, including SynthID, which Google began verifying in Search during May 2026, and then the harder half: C2PA Content Credentials handling as the standard broadens.
Why this work exists
It is the kind of thing that gets a line in a plan and no time in a calendar.
Part of Brand and Creative, and the part where the diagnosis is usually most useful on its own.
The nearest piece of work to it is finding the category convention worth breaking and the ones worth keeping. Where that is already handled, this moves faster than the estimate; where it is not, the two get done together or neither of them holds.
What good looks like
It is right when the next person to touch it does not have to start by working out what was done last time.
The gap between that and where most accounts start is why this is scoped as content provenance and AI disclosure rather than as a line inside something larger. Getting the photography to look like the business rather than a stock library is the usual second step once it is in place.
When this is the right work
The symptoms that point at this rather than something else:
- You use generative tools for imagery and have no policy
- Nobody can say which published assets were AI-generated
- A client or a regulator has started asking
If none of those is true, something else is probably the constraint, and the Brand and Creative overview maps the alternatives. The usual neighbours are making the brand legible in an AI answer where there is no design at all and deciding how much of the identity needs to survive a redesign.
What the work covers
What the work actually involves:
- Watermarking for AI-generated visuals, including SynthID, which Google began verifying in Search during May 2026, which is where most of the time goes and almost none of the budget.
- C2PA Content Credentials handling as the standard broadens, done before anything downstream is committed to.
- A disclosure policy that says who is responsible rather than which tool was used, done before anything downstream is committed to.
- Audit of existing assets, so you know what is already published and what it carries. The effort here scales with the size of the estate, not with the size of the brief.
How it runs
It runs in three passes. The order is not interchangeable.
Clearing the contradictions
Where two systems disagree about the same fact, that gets settled before anything is built on top of either of them.
Setting the review rhythm
How often this gets looked at again, and by whom, is agreed now rather than left to whoever notices first.
Costing the options
Where there is more than one route, each gets a rough cost and a rough timeline, so the choice is made on trade-offs rather than on whichever was described most confidently.
What each part runs into
Each piece has its own friction:
- Watermarking for AI-generated visuals, including SynthID, which Google began verifying in Search during May 2026, which in Indore runs into a reporting line that averages away the market in question
- C2PA Content Credentials handling as the standard broadens: the London version of this has to deal with a market where buyers qualify by area rather than by city
- A disclosure policy that says who is responsible rather than which tool was used, where a Los Angeles engagement contends with a competitive set that changes street by street
- Audit of existing assets, so you know what is already published and what it carries – working in Toronto this means accounting for a decision made by somebody who never visits the site
Where people lose time on this
Waiting for a rule. The verification is already live in Search, and retrofitting provenance to a library of published assets is far harder than doing it at production.
It is worth stating because it is the version of this work that gets sold most often, and it fails in a way that looks like the discipline not working rather than the execution being wrong.
How it is measured
Share of generated assets carrying credentials, and policy coverage across teams. That measure is agreed before the work starts, so there is no argument about the result afterwards.
Agreed before the work starts, so there is no argument about the result afterwards.
What you end up with
What arrives at the end of it: a written policy, a watermarking workflow, and an inventory of what is already out there.
Handed over with the working, not only the conclusion.
How it differs by sector
The same piece of work changes shape depending on the sector it is done for:
- In Automotive the binding constraint is that Local inventory, model-level demand, and a buyer who arrives having already decided.
- Healthcare and clinics: the thing that changes the scope here is that content that affects health decisions is held to a higher standard, and advertising is restricted.
- For Manufacturing and industrial, low volume, high value, and a specification sheet that decides the deal. and that shapes how far this can go.
The industry pages carry the rest of that detail. Automotive and Healthcare and clinics are the two where this piece of work most often decides the outcome.
Common questions
Do we have to disclose AI-generated images?
Increasingly the asset discloses itself: SynthID travels with the file and Google verifies it in Search. A stated policy is better than being asked without one. The specifics live in Agentic Commerce in Birmingham. AI Automation and Workflows in Birmingham is the context that makes them make sense.
Does disclosure hurt performance?
No evidence that it does. Being caught without one is the risk worth managing.
Is the output a report or a change?
Both, in that order. A written policy, a watermarking workflow, and an inventory of what is already out there. The report exists so the change is accountable, not instead of it. Migration and replatform planning goes through this properly, and Structured data implementation shows where it stops working.
How do we know this is the right piece of work?
The signals above are the test. If none of them describes your situation, something else is the constraint, and saying so is more useful than selling this. Start at Ahmedabad if nothing has been built yet, Ajmer if something has and it is not working.
What happens if it does not work?
We report that it did not, with the reason. Share of generated assets carrying credentials, and policy coverage across teams agreed before the work starts makes that an honest conversation rather than an argument. Ecommerce and retail is where the measurement is defined, Education and training where it gets disputed.
Ask about content provenance and AI disclosure
A sentence about the problem beats a filled-in form. We read every one of these ourselves.
Ask what we would do first, and why
Thirty minutes, the data you already hold, and a straight answer about where the constraint actually is.