Brands keep writing AI usage policies, but most teams can’t trace the files they use every day. PageProof CTO Marcus Radich explains why provenance checks are becoming a new baseline.
Creative teams spent the last year racing to build AI guardrails. They set rules for when synthetic media is allowed, drafted disclosure guidelines, and debated the boundaries of experimentation. Yet inside most workflows, a quieter problem has taken hold. Teams often have no reliable way to confirm the origins of the assets they use. A designer may receive a photo from a contractor. A motion artist may inherit a clip from a previous campaign. An influencer may hand over a piece of content that looks polished but arrives without context. If teams cannot trace where something came from, no amount of policy writing will protect them.
Marcus Radich, CTO of PageProof, sees the gap widening as AI-generated imagery, audio and video become routine rather than remarkable. To deal with that reality, his team recently launched FinalFile™, a free tool that allows anyone to verify and sign content credentials through a simple drag-and-drop workflow. He describes credentials as a quiet but essential marker inside a workflow. They tell teams who created a file, what tool generated it, and whether it has been altered. “It is tricky for the average person to sign their work,” he says. “You look at it and think, how do I get my content credentials on my work? It is not trivial.” Making that process simple is the point. If creative teams cannot verify what they are using, they cannot judge whether it aligns with company policy.
Provenance checks also reshape the production process long before a campaign goes live. Radich says the more crowded the creative pipeline becomes, the more important early checks become. “In the past you would take the finished artwork and go, fantastic, love it, let’s show the customer,” he says. “These days it would be a good practice to check the content credentials and see where it was generated, how it was done.” He has not yet seen widespread public failures where AI sourced assets accidentally shipped in major campaigns, but he believes the pattern is already forming behind the scenes. The volume of synthetic media is rising and creative handoffs are only getting more complicated.
Radich also points to the way the pattern of trust online has flipped, which makes verification more important than ever. “Today I distrust everything I see until proven true,” he says. “I flipped. I used to trust what I saw until proven not true. Now it is the opposite.” He describes a reversal from the early internet, when images and videos were accepted as real until proven otherwise, a change driven by how quickly synthetic media can be produced and shared. He worries that as false visuals circulate widely, they begin to shape how people remember events. “As time goes on, the memory of an event is represented by the media you can get access to,” he says. Provenance gives creative teams something objective to check as they build work for audiences who now bring a new level of skepticism to everything they see.
The question is not just what teams produce but how they protect their own creators. Influencers and content partners often build assets that are lifted, reposted and repurposed without attribution. Radich believes signing work can help them assert authorship and maintain value. “Signing their posts would be a good starting point to literally say, this is my original content,” he explains. Without that layer, the work is vulnerable to being copied without credit. “If you do not sign it, there is no way you are going to be credited for it.” The same logic applies to brand assets. AI models scrape anything they can reach, and unsigned work offers no signal about ownership.
His broader concern is accuracy. Radich has watched synthetic media reshape how people understand events and individuals, creating a distorted record that can be repeated and amplified by the same models that produced it. He calls this the “rewriting of history” and it drives his personal approach to verification. “If I am doing anything these days, I use three models and a search to verify a thing that I am working on,” he says. “I never just do it with one and go, okay, that is possible.” Creative teams may not operate with the same level of caution, but the principle remains. Verification should not be optional.
Advertisement
Radich sees provenance checks becoming a standard part of creative operations, but only if teams have tools that fit into the natural rhythm of production. That gap is why PageProof built FinalFile™ as a separate service rather than a feature tucked inside its core platform. The idea was to create something lightweight and accessible enough that anyone could verify or sign content credentials without extra software or training. It is not the only solution emerging in the market, but it reflects the kind of simple step teams will need as AI generated images, audio and video move through more collaborators. The goal is to make verification an everyday habit, not a last minute chore reserved for sensitive projects and Radich believes the industry is heading toward that expectation.
FinalFile™ is one of the features inside PageProof’s annual 12 Days of Christmas release sprint, a tradition that highlights the company’s fast release culture. New updates roll out weekly and sometimes daily and this year’s slate includes expanded language support and deeper integration with Asana. Radich says features are chosen for December because they offer clear value while still being simple to adopt. FinalFile™ fits that approach. It requires no training and no additional software, which means provenance checks can happen early and often.
PageProof plans to integrate signing directly into its approval workflow in the future, turning provenance into an automatic step instead of a manual one. For now the company wants teams to build the habit by checking assets when they enter the pipeline and signing them once they are ready to ship. Radich believes the correction does not have to be complicated if teams start with simple steps that fit their natural process. As he puts it, “At least we know where it originated.”