AI has made it much easier to create content. Images can be generated in seconds, copy adapted for different markets, videos translated and product content repurposed at a scale that simply wasn’t possible a few years ago.
And right now, there is another reason why organisations need to pay closer attention to what happens after that content is created.
Since 2 August 2026, the transparency provisions in Article 50 of the EU AI Act have started to apply. They introduce new requirements around identifying and disclosing certain types of AI-generated and AI-manipulated content, depending on how the content is created and used.
So this feels like a good time to look beyond the regulation itself.
Because the practical challenge is bigger than compliance.
For many organisations, the real question is becoming:
Do we actually know which content was created or modified by AI, which version is approved, and what we are allowed to do with it?
That is where the challenge is starting to shift.
It is no longer only about how to use AI to create more content. It is also about how to stay in control once AI becomes part of normal content workflows.
Which assets were generated by AI? Which were only edited or assisted by it? Which version has been reviewed and approved? What can be published, where can it be used, and does that information stay connected to the asset when it moves into another system or channel?
These sound like fairly basic questions.
In practice, they are not always that easy to answer.
And trust is already becoming one of the barriers to scaling AI. In a recent Forrester poll, 29% of AI decision-makers said trust was the biggest barrier to generative AI adoption in their organisation. You can read the full article here.
So the AI challenge is quickly becoming a content governance challenge too.
At its simplest, AI content governance is about knowing how AI has been involved in content and what should happen to that content next.
That can include information such as:
The goal is not simply to label AI content.
It is to make sure people and systems have enough information to decide whether content can be trusted and used.
We have previously argued that behind every successful AI strategy is a smarter content strategy.
AI can only work reliably with the content and context it has access to. That makes lifecycle management, metadata and governance part of the AI foundation, rather than separate DAM concerns.
But strategy is only one part of it.
Once AI becomes part of real content operations, organisations also need practical rules for how it should be used. That was the focus of our AI Content Policy Guide: helping teams think through AI-generated and AI-assisted content, human review, approval, disclosure and traceability.
The Policy can be downloaded here.
The next challenge is making those rules work in practice.
A policy can say that AI-generated content should be reviewed before publication. But does the organisation actually know which assets were AI-generated?
A policy can say that certain content should be labelled. But can that information follow the asset from creation, through approval, into a portal, website or another channel?
A policy can restrict what AI systems are allowed to use. But can an AI assistant or agent actually tell the difference between the latest approved asset and an outdated one?
That is where policy meets content operations.
The policy can define the rules. The content infrastructure needs to make those rules possible to follow.
The AI Act puts more focus on transparency around AI-generated and AI-manipulated content.
But this does not simply mean that every asset touched by AI suddenly needs an AI label.
Article 50 includes different obligations for providers and deployers of AI systems. These include machine-readable marking of certain AI-generated or manipulated content and disclosure requirements in specific situations such as deepfakes and certain AI-generated public-interest content.
The requirements therefore depend on how AI has been used, the type of content and who is providing or deploying the AI system.
What the regulation does do is make a broader question much harder to ignore:
Do we actually know how AI has been involved in our content?
If that information is missing, inconsistent or only exists in someone's head, it becomes difficult to apply either regulatory requirements or your own internal policy consistently.
That is why we think the discussion needs to go beyond AI labelling.
There is also a bigger change happening around AI that has very little to do with regulation.
Scott Brinker recently described how AI tends to move bottlenecks rather than simply remove them. When it becomes easier and cheaper to create more content, build more features, launch more workflows and experiment with more AI, the challenge starts moving elsewhere.
Towards relevance. Prioritisation. Context. And increasingly, coherence.
That is an interesting way to look at content operations too.
For years, production has been one of the constraints. Creating another image, translating another piece of content or producing another market variation required time and resources.
AI changes that.
Suddenly, producing ten, one hundred or even one thousand variations is technically possible.
But that doesn't automatically make those variations useful.
Someone, or increasingly something, still needs to understand what they are, what they belong to, which market they apply to, whether they are approved and where they should go.
AI makes it easier to create more content. The harder job is keeping that content usable, trusted and connected.
We see the same issue from another direction in product content.
Many organisations already have the systems they need. Manufacturers, for example, may have a PIM for product information, a DAM for media, an ERP for commercial information, a CMS for websites and separate portals or systems for dealers, service organisations and customers.
In research highlighted by Ntara, 68% of the industrial manufacturing leaders surveyed had a PIM and 58% had a DAM.
Yet having both systems does not automatically solve the problem.
Because customers and employees do not experience those systems separately.
They experience the information that comes out of them.
A dealer needs the right product data together with the right image and documentation. A service technician needs information connected to the correct product or serial number. A website needs approved content for the right market. And increasingly, an AI system may need to consume information from several of these sources at once.
So the problem often isn't:
“We need another place to store content.”
It is:
“How do we make sure the right content, with the right context, moves between all of these systems?”
AI makes that gap more visible.
People are no longer the only ones creating, finding and consuming enterprise content. AI systems and agents are beginning to do it too.
And they depend on those connections being understandable.
A natural first step is to make AI involvement part of an asset's metadata.
For example:
That is already useful for the person working with the asset.
But it becomes much more valuable once it is structured and can be used throughout the content lifecycle.
If the information is structured rather than hidden in a comment, file name or someone's memory, you can search for it, filter on it, build workflows around it and use it to control how content is distributed.
So a simple AI label may be what the user sees.
Behind that label, there needs to be useful context that systems can understand.
Some of that context can come from the content itself.
Standards such as C2PA Content Credentials provide a way to attach provenance information to digital assets. That information can help organisations understand where an asset came from, which tools or processes were involved and what has happened to it over time.
Provenance is useful, but it is important to be clear about what it does and does not tell us.
It does not automatically tell an organisation whether the asset is correct, approved, rights-cleared or suitable for a specific market or channel.
Those are still business and governance decisions.
What provenance can do is add another trustworthy piece of information that can be combined with the context the organisation already manages around the asset.
That is where DAM starts to become particularly relevant.
DAM has traditionally helped organisations answer questions like:
AI does not make those questions less important.
Quite the opposite.
Now we also need to understand whether AI was involved, what provenance information is available, whether a human has reviewed the result and whether that content is suitable for a particular workflow or channel.
And there is another development coming quickly behind this.
Enterprise AI is moving from assistants that mainly find or generate information towards agents that can actually take actions.
At that point, access to content isn't enough.
An AI agent also needs enough context to know what it should use.
For example:
That is quite a different problem from simply giving AI access to a folder full of assets.
It needs content together with enough trusted context to make sensible decisions.
And this brings us back to coherence.
The more content, systems, channels and AI we add, the more important it becomes that the information around the content stays connected and understandable.
This is also influencing how we think about the continued development of QBank.
A lot of the foundation is already there.
QBank uses structured metadata to give assets context and supports permissions, versions, approval processes, rights information and controlled distribution. Metadata has always been central to this, not just for search and findability, but also for automation, integration and governance.
We are now extending that foundation further.
One area is making information about AI involvement easier to capture and understand, including the distinction between AI-generated and AI-assisted content.
Another is provenance. Where Content Credentials or other embedded provenance information is available, that information should be possible to read and connect to the metadata and governance around the asset.
And it also needs to be visible.
A user browsing a library or content portal should not have to open several metadata fields just to understand something important about an asset. AI status can be shown clearly, while richer information stays available behind it.
From there, the interesting part is what you can do with that information.
For example, it could be used to:
Some of this builds directly on what QBank already does today. Other parts are being developed or explored as standards, regulation and customer needs continue to evolve.
But the foundation is the same: structured, governed content with useful context attached to it.
Explore more in our blog: AI Content Governance starts in the DAM. Not the AI Tool.
We have talked before about making content AI-ready.
Usually, that means making sure content is structured, searchable, governed and rich enough in metadata for AI systems to understand what they are working with.
That still matters.
But AI readiness is starting to mean something more.
It is no longer enough for an AI system to understand what an asset is.
Increasingly, it also needs to understand:
Where did it come from?
Is this the right version?
Has it been reviewed?
Am I allowed to use it?
What happens if I publish it?
This is why we believe DAM has an important role to play in enterprise AI.
Not because every DAM feature needs an AI layer.
And not because DAM should become the AI platform.
But because people, systems, channels and AI all need access to the same trusted context around enterprise content.
A PIM may know what the product is. A CMS knows where a page is published. An ERP knows commercial and operational information.
DAM has an important role in knowing which content represents that product, its status and rights, and where and how it can be used.
The value comes when those systems work together.
There is a fairly natural progression here.
An AI strategy helps define where AI should create value.
An AI policy helps define how the organisation wants AI to be used.
An AI-ready content foundation gives AI useful, structured content to work with.
And content governance and provenance help determine whether that content can actually be trusted and used.
The final piece is activation: making sure the right content can move into the right workflow, system, channel or AI use case.
That is where we think content operations are heading.
AI will continue to make it easier to create more content. More systems and agents will also be able to consume and act on it.
The organisations that handle that well won't necessarily be the ones creating the most.
They will be the ones that can keep the content, context and rules connected as it moves.
Because creating content is becoming easier. Knowing what you can trust, what you can use and what you can safely put into action is becoming the harder part.
No. Article 50 contains different transparency obligations depending on the type of AI system, content and use case. Some obligations relate to machine-readable marking by AI providers, while others relate to disclosure by deployers in specific situations.
AI content governance is the process of keeping information about AI involvement, provenance, review, approval, rights and disclosure connected to content throughout its lifecycle.
AI-generated content is primarily created using AI, while AI-assisted content involves AI as part of editing, enhancement or another step in the creation process. Organisations may need different policies and workflows for the two.
C2PA is an open technical standard for attaching verifiable provenance information, often called Content Credentials, to digital content. It can help show where content came from and how it has changed.
A DAM can connect AI status and provenance information with metadata, permissions, rights, versions, approvals and distribution rules. This helps people — and potentially AI systems — understand which content can be used and under what conditions.
Want to make your content foundation more AI-ready?
Read Behind every successful AI strategy is a smarter ontent strategy.
Working on AI governance?
Our AI Content Policy Guide helps teams define practical rules around AI-generated and AI-assisted content, approval, disclosure and traceability.
Want to go deeper into DAM and AI governance?
Read AI Content Governance starts in the DAM. Not the AI Tool.