Analysis: Journalism After AI Still Needs Something Machines Cannot Provide

Artificial intelligence did not create journalism’s ethical problems. Journalists struggled with accuracy, bias, manipulation, conflicts of interest and commercial pressure long before generative AI appeared. Artificial intelligence has changed the scale and speed at which those problems can now develop.

Lower barriers, higher stakes

A journalist once needed time to write an inaccurate story. A photograph had to be altered manually. Producing convincing fabricated audio or video required technical knowledge and considerable effort. Generative AI has dramatically lowered those barriers. Text, photographs, voices and increasingly convincing video can now be created in seconds. The same technology can help journalists examine documents, organize information, translate material and accelerate research. That makes AI neither inherently good nor inherently dangerous for journalism. The ethical question is who remains responsible.

What the Associated Press actually says

The Associated Press has taken a particularly clear position, most recently reaffirmed in updated newsroom standards released in July 2026. AP allows journalists to use AI to draft headlines, summarize documents, and handle transcription and translation, but draws a firm line around reporting, sourcing, editorial judgment and verification, which remain the responsibility of human staff. Any output from a generative AI tool should be treated as unvetted source material, according to AP’s standards, and AP prohibits generative AI from adding or removing elements from its news photographs, video or audio. Amanda Barrett, AP’s vice president for standards and inclusion, put the underlying principle plainly when the organization first introduced its AI guidance: AP journalists are responsible for the accuracy and fairness of the information we share. That principle should become fundamental across journalism. AI can assist a journalist. It cannot assume editorial responsibility.

The temptation newsrooms face

This distinction becomes increasingly important as newsrooms face economic pressure to produce more material with fewer people. Artificial intelligence offers an obvious temptation. If a system can summarize a report, produce a headline, rewrite a press release and generate an illustration within seconds, why should a newsroom spend substantially more time doing those things manually? The answer is that journalism is not simply the production of sentences. Journalism is a process of deciding what information deserves attention, determining whether that information is reliable, finding what is missing, challenging interested parties and accepting responsibility when something is wrong. Those are editorial acts, and none can be delegated to a tool with no stake in getting them right.

When oversight becomes superficial

Research from the Reuters Institute for the Study of Journalism shows that news organizations are increasingly moving beyond informal AI guidelines toward formal governance structures. Some organizations have established dedicated responsible AI teams, while others require human review of AI-assisted material. The research also identifies a growing problem: human oversight can become superficial when journalists are expected to review too much machine-generated material. That is an important warning. Putting a human name at the end of an automated production process does not automatically create human journalism. Meaningful oversight requires enough time for a journalist to question the information, inspect the sources and independently decide whether the material should be published.

Disclosure is necessary, not sufficient

Transparency matters too. Research and newsroom experience increasingly indicate that audiences want to know when artificial intelligence has played a meaningful role in producing journalism, and a public AI policy can become part of the relationship between a publication and its readers rather than simply an internal technology document. But disclosure alone is not enough. A newsroom cannot publish inaccurate information and solve the ethical problem by placing an AI label beside it. Verification must come before transparency.

The harder problem is visual

The visual challenge may be even greater than the textual one. A fabricated photograph of a bombing, demonstration or political event can resemble documentary photography while depicting something that never happened. Once circulated through social media, the distinction between illustration and evidence can disappear almost immediately. That is why generated editorial illustrations should never be presented in a way that could reasonably lead readers to believe they are photographs of actual events; NordoVista applies that standard to its own use of illustrations. News organizations also need to reconsider how they protect confidential information. Uploading unpublished documents, source identities, interview transcripts or sensitive material into external AI systems can create real privacy and security risks, which is why AP specifically cautions its own journalists against placing confidential or sensitive information into AI tools.

Trust, not speed, is the real advantage

Another transformation is happening entirely outside newsrooms. People are increasingly asking AI systems directly for news. The Reuters Institute’s 2026 Digital News Report notes that use of generative AI tools has risen rapidly, with a small but growing share of people using chatbots to obtain news and information directly. That changes the relationship between journalists, platforms and audiences. A reader may receive information extracted from several news organizations without visiting any of them, and the original reporting can become invisible.

In contrast, an automated summary becomes the reader’s primary version of events. This makes professional verification more valuable, not less. The defining advantage of journalism in the AI era cannot simply be speed, because machines will almost always win that competition. It must be trust. Readers need to know that somebody checked the document, questioned the claim, examined the photograph, contacted the other side, and decided that the information met a standard before publication. AI can participate in that process. It should never become the reason the process disappears.

The most important ethical rule for journalism in the age of artificial intelligence may therefore be remarkably traditional. Technology can change how journalists work. It cannot change who is responsible for the journalism.