Two wire services keep photo authenticity off-limits to AI

The writer's problem

A wire-service photo editor receiving footage from an AI-assisted source, or a reporter tempted to let a chatbot draft a paragraph on deadline, works inside newsroom rules that most readers never see. The Associated Press and Reuters have each published their own standards for generative AI, useful to any writer or editor trying to understand where the line between infrastructure and journalism sits inside institutions built on documented fact.

What the documents show

AP's standards page, bylined 15 August 2023, states that 'the central role of the AP journalist... will not change' and that AP staff 'do not use it to create publishable content,' treating any generative-AI output as 'unvetted source material' requiring ordinary editorial judgment and sourcing standards. On imagery specifically, the page states AP does 'not allow the use of generative AI to add or subtract any elements' from photos, video or audio, and will not transmit AI-generated images 'suspected or proven to be false depictions of reality,' with a stated exception for an AI-generated illustration that is itself the subject of a news story, provided it is 'clearly labeled as such in the caption.' Reuters' journalistic standards take a parallel but separately worded position: 'the use of any AI-generated or AI-modified visual elements is strictly prohibited,' while narrower tools that do not alter underlying visuals, describing a scene or drafting a shotlist, are permitted. Reuters adds that it 'remains responsible for the content we publish, whether an AI tool is involved in its creation or not,' and instructs staff to treat AI-generated facts as unverified until independently checked.

The editorial choice

Both organizations draw the same underlying line, that AI may assist production tasks but may not originate or alter what is presented as documentary fact, while wording the exceptions differently. This is an editorial position both wire services state explicitly: labeling an AI-generated illustration as such, in AP's case, is treated as compatible with editorial integrity precisely because the caption tells the reader what they are looking at, whereas an unlabeled or photographic-looking AI image is not.

What stays with the author

Neither organization's page specifies how a freelance contributor's AI use is verified before filing, or what happens when AI-generated material arrives already embedded in wire-fed content from another outlet; both standards place the burden on staff judgment and source-checking rather than on any automated detection system. The named responsibility, for accuracy, fairness and authenticity of what is published, sits with the journalist and editor in both documents, not with the tool.

• Is the AI tool being used to produce publishable content, or only as unvetted source material subject to normal verification?
• If an image is AI-generated, is that fact disclosed in the caption or credit, as both standards require for the narrow cases they permit?
• Does incoming wire or freelance material carry any indication of AI involvement this newsroom's own standards would require disclosing?

The two organizations reach similar conclusions independently, which is itself useful evidence for any writer assessing whether a given AI use is inside or outside a defensible editorial line.