The 30-second version. A publisher’s asset problem is rights, provenance and deadline at once. Every editorial image carries a caption, a credit or byline, and a usage restriction — licensed, wire, contributor, embargoed — that a legal or ethical obligation depends on, and it all moves under deadline pressure. A DAM’s saving here is 100% IPTC/XMP metadata fidelity so captions, credits and rights never get stripped; rights tracking so nothing runs outside its licence or past its expiry; and fast ingest and search so the right frame is found on deadline, not after it. The deeper reason is one failure repeating at three time scales: the obligation gets separated from the image — in seconds when a file is rescaled, in months when it passes through a platform, and across decades until the archive is full of orphans nobody can clear. The chain, documented →
Media & publishing has no dedicated ranking on this site, but the capabilities it lives on are directly tested elsewhere: this page cross-links the metadata-fidelity ranking for the editorial-metadata requirement and the photographer ranking for the deadline-culling and wire-captioning side.
The asset problem in media & publishing
Most industries treat metadata as an aid to search. In editorial, the metadata is the obligation. A single news or feature photo carries a caption that must be accurate, a credit or byline that must be attached, and a usage restriction that must be honoured — and those obligations are legal and ethical, not merely tidy. IPTC, the standard our whole metadata testing is built on, is literally the International Press Telecommunications Council’s; as our glossary puts it, for any library handling licensed stock, contributor content, or press assets with usage restrictions, IPTC fields are often the only reliable record of who owns what.
Two things make that expensive. Editorial libraries are rights-dense: licensed stock has a usage scope and an expiry, a contributor’s photo may be cleared for one story and not another, wire images come with strict terms, some assets are embargoed until a date. Publish outside any of those and it is a correction or a legal exposure. And it all runs on deadline: photographers cull thousands of frames against the clock, and an editor who can’t find the right approved, cleared image in time either misses the slot or runs the wrong one.
What editorial teams actually struggle with
There is one failure under most of this, and it repeats at three time scales: the obligation gets separated from the image — in seconds when a file is rescaled, in months when it passes through a platform, and across decades when an archive outlives the people who licensed it. The US Copyright Office named the endgame of that drift in its 2015 orphan-works report: photographs dominate the problem precisely because they “frequently lack or may become divorced from ownership information.” Everything below is a link in that chain, cited to primary documents in sources.
1. The obligation travels in a free-text field
IPTC — the standard editorial actually runs on — has the right vocabulary. Credit Line carries the credit the supplier requires; Copyright Notice carries the claim; Licensor names who to contact; Instructions is where embargoes are conveyed; and Rights Usage Terms holds the licensing parameters. The catch is in the standard’s own definition: Rights Usage Terms is free text. It is prose for a human, not a rule a machine can enforce. No system can compute “this expires in March” from a sentence. That is precisely why rights have to be modelled as structured fields with real dates in a catalogue — the embedded string is a note, not a control.
2. Stripping the credit is its own legal exposure
Most teams treat a lost credit as sloppiness. In the US it is a separate statutory question. Copyright law protects “copyright management information” — and the definition expressly includes the author’s identification, the owner’s details, and the terms and conditions for use of the work. Removing or altering it is addressed on its own, independent of whether the underlying use infringes.
Where we stop short: the provision turns on intentional removal and a knowledge element, and courts have not settled whether an automated pipeline that rescales images meets it. Treat this as exposure to argue about, not a violation you can assume — the point for a publisher is that the credit field is legally freighted, not that every resize is actionable.
3. The platforms strip it anyway
IPTC has tested this directly, and the results are unkind: in its platform testing, some major social networks removed embedded metadata entirely, another stripped XMP specifically, several large free-photo sites returned downloads with all metadata gone, and mainstream cloud drives dropped metadata from preview-saved copies. The mechanism is the same one that bites in real estate: the metadata survives while the pixels are untouched and dies on rescale — and rescaling is what publishing pipelines do all day.
Date this carefully: those IPTC results are from 2019. Platforms change. Do not quote them as current behaviour — and don’t let a vendor quote them at you either. Test your own pipeline: push a file with full Credit Line, Copyright Notice and Rights Usage Terms through your CMS, pull the published derivative back, and read the fields.
4. “Editorial use only” is narrower than most newsrooms think
The boundary isn’t just “don’t put it in an ad.” Getty’s licence terms bar editorial-marked content from commercial, promotional, advertorial, endorsement, gambling and merchandising uses — and the named inclusion of advertorial and endorsement is the one that catches editorial teams, because sponsored content and branded partnerships are exactly where an editorial image drifts across the line without anyone deciding to cross it. The same terms allow cropping and technical-quality edits but not alteration beyond that.
Note that agency terms vary by territory and by negotiated contract — read your own. The DAM answer is to hold the licence class as enforced metadata, so an editorial-only asset cannot be pulled into a commercial or sponsored workflow by someone who never saw the contract.
5. The credit has a required format, not just a requirement
Agencies specify the shape of the credit line and where it must appear — Getty, for instance, prescribes a photographer/collection construction and requires it adjacent to the content or in production credits. So “we credited them” isn’t automatically compliance; the wrong construction, or the right one in the wrong place, is still a breach. That is a templating and export problem: the credit should be assembled from fields, not retyped by whoever places the image.
6. Embargoes are two different mechanisms, guarded by upper case
Wire practice distinguishes a transmission embargo (nothing goes to any client until the time) from a publication embargo (the material is transmitted to media clients immediately, but they are restricted from publishing until the time). Those are different risks with different blast radii, and a lot of DAM copy collapses them into one word.
What’s striking is the mitigation. In Reuters’ handbook, staff moving embargoed material to an editing desk are told to mark it as embargoed in two places, in upper case — a human convention guarding a commercially explosive rule. The handbook even has a standing procedure for accidental release, and its instruction is telling: the material must not be deleted. You cannot un-send; you can only prove what went out and when. Read as requirements, those are scheduled release with hard gating and an audit trail that survives the incident — two things a folder cannot do and upper case cannot guarantee.
7. The manipulation line is absolute — and now includes AI provenance
Editorial image ethics are stricter than most industries’ compliance rules. Reuters’ handbook makes material alteration of a picture a dismissal-level offence, bars cloning and healing tools outright with a single exception for sensor dust, forbids staging or re-enacting news events, and requires that composites showing an event’s progression state the technique in the caption — while never being acceptable in a news assignment at all. Captions carry their own duties: source contentious figures, explain the circumstances, state the correct date, and describe only what was witnessed rather than what was assumed.
Cite this one carefully. The widely-circulated copy of the Reuters handbook is the second online edition, April 2008 (last modified 2009) and is hosted by a third party; its technical details are visibly of that era. The ethical principles are quotable and stable; the specifics are not current practice. We date it rather than imply it is today’s standard — and we cite no AP, BBC, Guardian or NPPA text here at all, because we could not retrieve those documents to verify them.
AI is where this gets newly concrete. IPTC’s Digital Source Type vocabulary now gives editorial a machine-readable way to say how an image came to exist — distinct values for content captured by a camera, content edited by humans with non-generative tools, content created by a trained AI model (trainedAlgorithmicMedia), a composite that includes generative AI, and generative augmentation such as inpainting. IPTC recommended in 2023 that AI-image software stamp that value, and the current spec adds fields for the prompt, the prompt’s author and the AI system used. Provenance is now a metadata field — which makes it a DAM responsibility rather than an editorial debate.
8. “We licensed it for the issue” is not forever
The most expensive publisher misconception has a Supreme Court answer. In New York Times Co. v. Tasini, freelance contributions were placed into electronic databases, and the Court held the collective-works privilege did not authorize it — the databases reproduced the contributions in isolation rather than as part of that particular collective work or a revision of it. It is a text case, but the logic governs photo archives exactly: a licence into an issue is not a licence into everything the issue later becomes.
Wire practice already reflects this. Reuters’ guidance on buying amateur and freelance material stresses using the appropriate forms to secure access both for news and for archive purposes — archive rights treated as a right you acquire deliberately, not one you inherit. If your archive was built without that discipline, the DAM job is to record, per asset, what was actually bought.
9. Rights rot, and the orphan-photograph problem
Run the drift for thirty years and you get orphans. The US Copyright Office found that photographs make up a significant share — if not the majority — of the orphan-works problem, and gave the reason plainly: photographs frequently lack ownership information or become divorced from it, leaving would-be users without even a starting point for a search. It also observed that the sheer volume of photographs and the absence of centralized rights-clearance machinery block licensing at scale.
That is the DAM thesis stated by a regulator, and it is the same sentence as struggles #1, #2 and #3 — rights information separated from the asset — only at the decade scale, where it becomes unfixable. The archive you can still clear in 2040 is the one whose rights were structured fields in 2026.
Not law: that report also floated remedy limitations for good-faith users and extended collective licensing for photographs. Those were recommendations in a 2015 report, not enacted US law — they are sometimes described online as though they were.
Where a DAM saves money here
- Metadata fidelity that protects captions, credits and rights. A tool that strips structured fields on export means — as we found testing round-trip fidelity — captions, rights data and creator contact are gone the moment you try to leave. For editorial, 100% IPTC/XMP metadata preservation isn’t a nicety; it is the difference between an archive you can trust and one that quietly loses attribution.
- Rights enforcement, not just rights storage. A DAM that can flag or block an asset whose licence has lapsed, or whose scope doesn’t cover this use, stops the out-of-licence republish before it happens — the point of rights management in a newsroom. An expiry sitting in an email protects nothing.
- Deadline speed — cull fast, find fast. Editorial shooters cull thousands of frames on deadline, and the fastest ingest tools exist for exactly this. A library where the right cleared frame surfaces in seconds, by facet or saved search, is the difference between making the slot and missing it.
- Standards-based portability. Because IPTC/XMP is written into the file, captions and rights travel with the image to any future system — you are never locked into one vendor’s database to keep your attribution.
How it plays out
An illustrative composite. The scenario below is not one named publication — it is a composite of the patterns we see, built entirely from capabilities and figures we have tested and published. No invented benchmarks.
Picture a publication with several feeds into one library: a wire service, staff photographers, freelance contributors, and licensed stock. Each image arrives with its own caption, credit and terms — and those terms differ per source.
On a shared drive, the licence expiry for a stock image lives in the purchasing email, so a striking photo gets republished months after the rights lapsed. A freelancer’s excellent frame runs with no credit because the byline never travelled with the file. An editor on deadline scrolls a folder of three thousand near-identical frames looking for the one that was actually cleared. Each of these is a routine, avoidable failure.
In a DAM, every image keeps its caption and credit embedded in IPTC/XMP, so attribution follows it everywhere; the usage rights and expiry live on the asset, and the tool flags the lapsed-licence photo before it runs again; and the cleared frame is a saved search away, not a scroll. The saving isn’t a percentage we can invent — it is the elimination of the mis-credit, the out-of-licence republish, and the deadline miss, all of which have real cost. Budget the one-time set-up honestly: cataloguing and tagging an existing archive runs about a week per 50,000 files, less with AI tagging.
The capabilities that matter most here
1. Metadata-fidelity round-trip
The editorial non-negotiable: 100% IPTC/XMP preservation so captions, credits, creator contact and rights survive export and re-import. Losing them is permanent data loss — our fidelity ranking measures exactly this across tools.
2. Rights that flag and block, not just store
Usage scope and expiry on the asset, with the tool able to flag or block an out-of-licence or expired image before it publishes. See rights management — a date nobody acts on protects nothing.
3. Deadline-grade ingest and search
Fast culling of thousands of frames and instant retrieval of the cleared one — via faceted filtering and saved searches. Speed on deadline is the difference between making the slot and missing it; the photographer ranking covers the culling side.
4. Standards-based, portable metadata
IPTC/XMP written into the file so attribution and rights travel with the image to any future tool — no lock-in to one vendor’s database to keep your credits. Covered under XMP, where a publisher can define custom fields too.
Buyer’s test: during a trial, import a photo with a full caption, credit and a usage-restriction field, export it, re-import it and diff the metadata — every editorial field should survive intact. Then set a licence expiry on an asset and confirm the tool actually flags or blocks it once past the date, rather than just storing a date nobody sees. A DAM that loses a credit on round-trip, or treats an expiry as inert text, is unsafe for editorial.
FAQ
Why do publishers and newsrooms need a DAM specifically?
Because editorial assets are rights-dense and deadline-driven in a way most industries aren't. Every image carries a caption, a credit and a usage restriction - licensed, wire, contributor, embargoed - that a legal or ethical obligation depends on, and it all moves on deadline. A shared drive can't guarantee those fields survive, can't stop an out-of-licence republish, and can't surface the one cleared frame fast enough to make the slot.
What's the single most important DAM capability for editorial?
Metadata fidelity. In editorial the metadata is the obligation - the caption, the credit, the rights - so a tool that strips structured fields on export is dangerous, not just inconvenient: attribution and licence terms are gone the moment you migrate. 100% IPTC/XMP round-trip is close to non-negotiable for any library handling licensed or press content.
How does a DAM prevent publishing a photo outside its licence?
By keeping the usage scope and expiry on the asset and acting on them - flagging or blocking an image whose licence has lapsed or whose terms don't cover the intended use, before it runs. That is the difference between rights management and rights storage: a licence expiry sitting in a purchasing email protects nothing, while a DAM that enforces it stops the correction before it happens.
Does a DAM help meet editorial deadlines?
That's a core benefit here. Editorial shooters cull thousands of frames against the clock, and the fastest ingest and search let an editor surface the right cleared, credited frame in seconds rather than scrolling a folder. A saved search for 'this story, cleared, unpublished' turns a deadline scramble into one click.
Is there a DAM ranking specifically for media and publishing?
Not on this site yet. The two closest reads are the metadata-fidelity ranking - the editorial requirement that captions, credits and rights survive - and the photographer ranking, which covers deadline culling and wire-service captioning.
Is 'editorial use only' just about not running the photo in an ad?
No - it is broader. Getty's licence terms bar editorial-marked content from commercial, promotional, advertorial, endorsement, gambling and merchandising use, and it is 'advertorial' and 'endorsement' that catch newsrooms: sponsored content and branded partnerships are exactly where an editorial image drifts across the line without anyone deciding to cross it. The same terms allow cropping and technical-quality edits but not alteration beyond that. Agency terms vary by territory and by negotiated contract, so read your own - and hold the licence class as enforced metadata rather than a note someone has to remember.
Is stripping a photo credit actually illegal?
It is at least a separate legal question from infringement. US copyright law protects copyright management information, and the definition expressly includes the author's identification, the owner's details and the terms and conditions for use of the work - so credit and rights fields are legally freighted, not merely tidy. The caveat matters: the provision turns on intentional removal and a knowledge element, and courts have not settled whether an automated pipeline that rescales images meets it. Treat it as exposure to argue about, not a violation you can assume.
How should a publisher label AI-generated or AI-edited images?
There is now a standard rather than a debate. IPTC's Digital Source Type vocabulary provides distinct machine-readable values for content captured by a camera, content edited by humans with non-generative tools, content created by a trained AI model, a composite that includes generative AI, and generative augmentation such as inpainting. IPTC recommended in 2023 that AI-image software stamp that value, and the current standard adds fields for the prompt, the prompt's author and the AI system used. Provenance is a metadata field now, which makes it a DAM responsibility to carry and re-stamp on export.
Does licensing a freelance photo for an issue cover the archive?
Not automatically. In New York Times Co. v. Tasini the Supreme Court held that the collective-works privilege did not authorize reproducing freelance contributions in electronic databases, because they appeared in isolation rather than as part of that particular collective work or a revision of it. It is a text case, but the logic governs photo archives: a licence into an issue is not a licence into everything the issue later becomes. Wire practice reflects this, treating archive rights as something acquired deliberately at the point of purchase.
Sources & references
- IPTC field definitions — IPTC Photo Metadata Standard 2025.1: Credit Line, Copyright Notice, Creator, Licensor, Description, Instructions (where embargoes are conveyed) and Rights Usage Terms. The standard itself defines Rights Usage Terms as free text — the basis for our point that embedded rights are a note, not a control.
- AI provenance. IPTC Digital Source Type NewsCodes (incl.
trainedAlgorithmicMedia, composite and human-edit values) and IPTC's 2023 guidance for AI-generated synthetic media. - Metadata stripping. IPTC's platform metadata tests. Dated 2019 — we cite the mechanism, not current per-platform behaviour, and recommend testing your own pipeline.
- Credit as protected information. 17 U.S.C. §1202 — copyright management information, whose definition includes the terms and conditions for use of the work. Requires intentional removal and a knowledge element; unsettled as applied to automated pipelines — we present it as exposure, not certainty.
- Editorial-use restrictions & credit format. Getty Images licence agreement — the editorial-only bar on commercial, promotional, advertorial, endorsement, gambling and merchandising use, and the prescribed credit construction and placement. Terms vary by territory and negotiated contract — read your own.
- Embargoes, manipulation and captions. Reuters Handbook of Journalism, 2nd online edition, April 2008 (third-party hosted). Explicitly dated: the ethical principles are stable, the technical specifics are of their era. This is not presented as Reuters' current standard. We cite no AP, BBC, Guardian or NPPA text anywhere on this page because we could not retrieve those documents to verify them.
- Freelance re-use. New York Times Co. v. Tasini, 533 U.S. 483 (2001) — the collective-works privilege does not extend to reproducing contributions in electronic databases.
- Rights rot. Orphan Works and Mass Digitization: A Report of the Register of Copyrights, US Copyright Office, June 2015 — photographs as the bulk of the orphan-works problem, and ownership information becoming divorced from the image. The report's remedy-limitation and extended-collective-licensing ideas are recommendations, not enacted law.
- IPTC — the International Press Telecommunications Council standard; "for any library handling licensed stock, contributor content, or press assets with usage restrictions, IPTC fields are often the only reliable record of who owns what."
- Metadata-fidelity ranking — captions, rights and creator contact lost the moment you leave a tool that strips fields on export; 100% round-trip as the archival requirement. July 2026.
- Photographer ranking — Photo Mechanic for "sports, news and event shooters who cull thousands of frames on deadline"; IPTC stationery as "the gold standard for wire-service captioning." July 2026.
- Rights management and XMP — usage scope and expiry on the asset; a publisher defining custom editorial fields; the "week per 50,000 files" tagging figure from the small-business ranking.
The editorial-metadata, rights and deadline-workflow capabilities are drawn from our testing and reviews; the composite case invents no publication and no figures, per how we source claims. See how we test.