It’s impossible not to notice that generative AI is the hot topic right now, and its explosive popularity means newsrooms of all types—especially those that broadcast images to the world—face authenticity challenges that go far beyond traditional issues of whether a certain clip is fake or doctored.
Newsrooms must now grapple with whether they can show how and where a particular clip came from, the process by which it reached the edit and whether the version that viewers see still matches the material it claims to represent.
In our increasingly low-trust media landscape, AI blows everything up at a time when many major broadcasters can scarcely afford more reputational harm and attacks, especially around the basics of authenticity and provenance.
A Delicate Balance
On both sides of the trust equation, the balance is becoming more delicate. Currently, AI tools can make scarily convincing images, video and audio in moments, a far cry from the stodgy, copyright-infringing outputs of even 12 months ago.
C2PA, or the Coalition for Content Provenance and Authenticity, is one attempt to shift that work earlier in the broadcast chain. The group, formed in 2021 and designed to protect content by building an open, industry-wide cryptographic standard to map the origin and history of digital media, features most of the heavy hitters in tech, including Google, Microsoft, Meta, OpenAI, Adobe and Amazon.
Rather than asking journalists, platforms or audiences to spot every possible fake after the fact, it creates a way to attach tamper-evident provenance information to media as it is captured, edited, distributed, and displayed.
To many broadcasters, the appeal of the C2PA solution should be obvious. In theory, a piece of video could carry a technical record of its journey from camera to newsroom, through the edit, out to publication, and eventually into a player or TV app.
In practice, thorny questions remain over whether that chain can survive the real broadcast workflow—from ingest to playout, streaming and social distribution—and the messy hand-offs that are a natural part of this process.
What C2PA Can Prove
At a basic level, C2PA is best understood as a way to preserve the history of a piece of media, rather than as a tool that decides if a story is true.
C2PA lets a validator check that a signed manifest belongs to a particular asset, whether the asset still matches that cryptographic binding and what the signer has asserted about its creation or editing history. It does not independently prove that every assertion is accurate, that nothing happened before the first signature or that the signer should be trusted.
In other words, C2PA is nowhere close to a panacea, and the most useful way to think about it is as a chain of custody.
As a Sinclair spokesperson put it: “Authentication and provenance answer questions such as: Where did this content come from? Who created it? Has it been altered? They do not independently determine whether every claim within the content is accurate or true.
“Editorial standards, fact-checking, reporting, and journalistic judgment remain essential,” the spokesperson said. “Provenance is a valuable tool that supports trust, but it does not replace the work of journalism.”
Bruce MacCormack, chair of the International Press Telecommunication Council (IPTC) Media Provenance Committee and co-founder of Project Origin, a consortium of global news organizations formed to combat digital disinformation by establishing a secure, end-to-end process to prove the integrity and origin of digital news and journalistic content, summed it up nicely: At both ends of the broadcast chain—ingest and output—knowing the provenance of media files is vital.
At ingest, signed content from a trusted camera, newsroom, agency or partner can help verification teams triage material more quickly; unsigned footage does not automatically become suspicious, but it may need extra manual checks.
“The issue is the increased fact-checking burden created by the wide availability of tools to generate or manipulate media,” MacCormack said. “C2PA can provide an ‘express lane’ for media that is signed by trusted parties.”
C2PA can provide an ‘express lane’ for media that is signed by trusted parties.”
Bruce McCormack, IPTC
At output, the same infrastructure can help a broadcaster defend its own material once it leaves the newsroom, especially when clips are reposted, embedded or shared on third-party platforms.
A valid signature is therefore only part of the calculation. Newsrooms also need to establish who signed the credential—a camera, application, agency or publisher—and whether that identity is trusted in the relevant context. A credential from an unknown signer can be cryptographically intact without carrying much editorial weight.
Signing at the Source
The cleanest and earliest place to start that chain is in the camera.
If a newsroom can verify a video or image file only after it has been edited, compressed or pulled from a platform, it is already fighting a losing battle. Camera-side authentication aims to move the first trust signal as close as possible to the point of capture.
For Sony, that means building provenance into the camera (or media capture device) itself.
According to Sony Spokesperson Allison Mandara, Sony signs the captured file inside the camera, binding the media to the device so that downstream systems can validate its origin and integrity.
C2PA-compliant credentials then provide a standard way to record how and when media was created. In Sony’s case, capture time can help establish when the material was recorded, while 3D depth data—where supported—can add evidence the image came from a physical camera capture, rather than a generated or artificially constructed scene.
Camera-side signing is only the first link, though. Mandara said provenance is strongest when it remains attached “from camera capture all the way through editing, storage, and distribution workflows.”
The challenge is that real-life broadcast media rarely travels in a straight line. “Provenance can be lost when content is exported, transcoded, compressed, screen-recorded or moved between systems that do not yet support these standards,” Mandara said. “That is why industry-wide adoption and workflow integration are critical. Authentication needs to become part of the normal production pipeline rather than an additional step.”
Sony is also trying to move verification out of standalone tools. Mandara said Camera Verify and its Image Verification Pre-Check SDK can be integrated into newsroom or media-management systems, allowing routine checks to run without requiring journalists to process every file manually.
Because news footage is rarely broadcast exactly as captured, the edit must become part of its provenance record.
Adobe Premiere and Media Encoder can attach Content Credentials at export, including verified identity and linked accounts. Premiere can also scan supported imported assets for existing credentials, while sequences containing supported Adobe AI-generated material receive credentials automatically in compatible export formats.
Adobe: Through the Edit
A clip may be trimmed, captioned, color-corrected, combined with other footage or prepared for several platforms before it reaches an audience—all of which are completely standard processes. Those changes do not make the media suspicious by default, but they do need to be carried out honestly if provenance is going to remain useful.
Each handoff still must understand what it is receiving, and cross-compatibility remains a potential stumbling block.
Adobe itself acknowledges that it cannot guarantee every publishing destination will support Content Credentials, which neatly captures the wider broadcast problem: one system can preserve provenance, but the chain only works if the next system does. too.
Any handoffs become more complicated inside a professional broadcast environment.
Much of the public conversation around C2PA has focused on capture, publishing and consumer-facing formats, but broadcasters work across a much messier chain of file formats, live signals, production tools, playout systems and distribution routes.
For Thomas Bause Mason, director of standards development at the Society of Motion Picture and Television Engineers, that means provenance has to be considered across the standards that already hold broadcast workflows together.
“SMPTE does not want to invent its own CPA specification but rather make sure that SMPTE standards are ready to support the CPA solution the market will choose,” he says. “That may be C2PA.”
In practice, the question is whether the systems used for professional editing, media exchange, IP production, legacy video links, and compression can carry a provenance record without losing it along the way.
The Material eXchange Format (MXF) is becoming more widely used for professional editing and distribution, but Bause Mason said no single format or handoff can solve the problem on its own. Broadcast workflows are made of many connected systems, and each one needs to know how to preserve the trust chain.
SMPTE does not want to invent its own CPA certification but rather make sure that SMPTE standards are ready to support the market will choose. That may be C2PA.”
Thomas Bause Mason, SMPTE
Bause Mason’s broader argument is that professional media infrastructure is not yet prepared to support content provenance end to end. Existing standards and systems need ways to carry manifests, identifiers or other durable links to provenance data before C2PA can move from controlled demonstrations into routine broadcast operations.
Editing, VFX, transcoding and format conversion can strip or break provenance, but they should also be able to add their own assertions to the manifest so the credential reflects what actually happened to the media.
Multiple Layers
Provenance also must represent several layers of origin. From a broadcaster’s perspective, the producer may be the source, but the audience may need to know which channel, platform or publisher endorsed the final version.
As one might imagine, this creates some difficult broadcast-specific questions.
Bause Mason helpfully frames some of them: Should authentication work frame by frame or shot by shot? How should provenance remain synchronized with the content during live production? What happens when contribution feeds from OB trucks are cut into a program, sent to consumer outlets and then passed to affiliates for further edits?
As he elegantly put it, “Only if the whole ecosystem is trusted glass to glass can the consumer be sure to receive content they as well can trust.”
When The Chain Breaks
A broken provenance chain cannot become a verdict in itself. If C2PA is going to work in a newsroom, it has to support editorial verification rather than replace it.
For broadcasters that handle user-generated video, archive footage, agency or wire material, social clips and content from third parties—in other words, many of them—this point is especially salient.
Some of the content may never have had credentials attached in the first place; some may have passed through systems that stripped or corrupted metadata before it reached the newsroom; and so on.
A newsroom may encounter media that was never signed, media whose manifest has been stripped, a credential that fails validation or a perfectly valid credential from an unfamiliar source—four different cases requiring different editorial responses.
As Marcos Armstrong, senior specialist, content provenance and integrity at CBC/Radio-Canada put it: “When dealing with user-generated content, archival assets, or third-party wire material that lacks a cryptographic signature, broadcasters cannot rely on data alone.”
At CBC/Radio-Canada, that means provenance sits alongside more traditional verification work. Armstrong points to visual vetting, geolocation, satellite imagery, public records and other open-source intelligence techniques used to check visual material when credentials are absent, incomplete or broken.
Armstrong said CBC/Radio-Canada is also exploring watermarking and perceptual fingerprinting as ways to reconnect broken credentials to their original manifests.
CBC/Radio-Canada has also tested more complete provenance workflows, including an in-house signing tool, the EBU’s C2PA-enabled video player, a C2PA-enabled Sony PXW-Z300 and Adobe Premiere in demonstrations of how a “glass-to-glass” workflow might operate.
Provenance can also fail for ordinary reasons. A genuine piece of footage may lose its credentials during compression or resizing, while a misleading video can still carry a clean technical chain.
For broadcasters, C2PA adds another signal to the verification process, helping newsrooms see where confidence is stronger, where the chain has gaps and where human judgment still needs to take the lead.
Reaching the Viewer
Even if provenance survives the production chain, it must still reach the audience in a form people can understand. Sinclair sees that display layer as central to adoption.
“For content authentication to become a routine part of broadcast news and production workflows, the standards work needs to move beyond carrying provenance in technical metadata and ensure that the end viewer can actually see and understand it,” a spokesperson for the station group said.
Authenticated enterprise identity needs to be shown consistently in media players, televisions, and software players, the broadcaster said. Without that, a piece of content may carry valid provenance data, but the viewer may never see that it came from Sinclair or another trusted source.
There is a UX problem here as well as a standards problem.
Armstrong said CBC/Radio-Canada has been exploring how Content Credentials should appear on its digital platforms without overwhelming users or making routine edits look suspicious, a delicate balance.
CBC’s proposed model uses progressive disclosure: an unobtrusive icon or banner first, followed by an expandable plain-English summary for users who want more detail. The interface would distinguish routine technical changes, such as resizing or colour-profile normalisation, from edits that can affect context, such as a heavy crop, and from transformative changes involving generative AI.
For provenance to become routine, it has to disappear into the workflow. C2PA only becomes useful to broadcasters at scale if it is built into cameras, editing tools, asset-management systems, CMS platforms, distribution chains and players.
From Trial to Broadcast Infrastructure
MacCormack argued that media provenance should become a feature rather than a standalone product—something built into capture, edit, distribution and display systems rather than handled by specialist signing or validation tools outside the normal workflow.
Armstrong makes a similar point from the broadcaster side, saying that the goal is to move away from manual resigning and embed provenance natively into common tools.
For Bause Mason, the standards challenge reaches across the professional media ecosystem, with each transformation point preserving the credential or identifier while adding its own record of what changed.
Broadcasters are not there yet. Workflows are mixed-vendor, format-heavy and full of hand-offs that were never designed to preserve cryptographic manifests. Social platforms, browsers, operating systems, televisions and software players also need to recognize and display provenance consistently before audiences can benefit from it.
C2PA will not replace newsroom judgment or settle every dispute over whether a piece of media is trustworthy. Its value is more practical: If the chain can hold, broadcasters gain a way to show where media came from, how it moved through production and why viewers can have confidence in the version they are watching.
