Anthropic just found traditional media’s biggest advantage and we need to pay attention

New Delhi: What if I told you that, in the age of AI, traditional media may be sitting on one of its biggest advantages and hardly anyone is talking about it? Television, radio, and print are frequently told that their best days are behind them. The audience is moving online. Digital can target people better, measure them better, and hence, advertising money too follows. Dentsu's 2026 India report shows digital accounted for only 12% of Indian advertising expenditure in 2016. By 2025, it had risen to 59%. WPP Media expects India's advertising market to touch Rs 2,01,891 crore in 2026, up 9.7%. But digital, including the digital extensions of other media, is already expected to take 68.1% of total ad revenue. The numbers make defending traditional media difficult. But an Anthropic investigation done on its own tool and its usage threw up an unexpected twist. And that might be the biggest differentiator the traditional news ecosystem needed. The fake elephant in the room The Claude parent did an investigation on how its GenAI platform is being used in the domain of influence and information. One operation linked to France-based digital advertising agency LKM Company operated around 70 fabricated news websites, another 70 matching X accounts and more than 250 inauthentic commenting accounts. The network published at least 8,913 articles in around 20 languages. Claude was used to write stories, rewrite genuine journalism with different political slants and create what looked like separate local newsrooms. Fake journalists signed the stories. AI-generated identities helped amplify them. Coordinated accounts commented underneath. In other words, Claude could manufacture almost every signal we normally associate with a piece of content being popular. Here’s the plot twist While the research drew this dystopian picture, it also found out an inconvenient truth. Most of these content pieces failed to garner genuine audiences. Despite all the infrastructure containing artificial publishers, journalists, and readers, the operation largely remained inside the ecosystem it had manufactured for itself. And now comes the most interesting observation Anthropic made in the report. After looking across the influence operations it uncovered, Anthropic said most content received “little or no authentic engagement.” But the widest authentic reach appeared when traditional media organisations provided the distribution including FM radio, satellite and shortwave radio, and television. Read that again. When these artificial influence operations wanted to reach genuine human beings, established and traditional media distribution was the only way. That is a remarkable finding at a time when traditional media is repeatedly told that its audience advantage has disappeared. Because perhaps we have been looking at the wrong advantage. Digital has scale, targeting and speed. Now, with AI, content supply on digital is almost infinite. But television, radio and established news brands still have something much harder to create overnight. A real audience that existed before the fake content arrived. Orwell’s 1984 To push the irony a little further, I am reminded of a moment in George Orwell’s 1984 that feels very relevant here. The protagonist in the book, Winston Smith, works at a place called the Ministry of Truth, where history is constantly rewritten to fit whatever the ruling Party wants people to believe today. While working there, he comes across an old newspaper clipping. The clipping places three disgraced Party leaders, Jones, Aaronson and Rutherford, at an event in New York on the exact date they were supposedly committing treason somewhere else. The tiny piece of evidence was explosive for Winston. That small clipping gave Winston the realisation of something real that once happened. That stubborn piece of paper pointed out that the official story can be rewritten but the physical record says something else happened. The internet today is obviously not Orwell's Oceania. But AI is creating a similar problem in a very different form. When almost every signal around information can be generated, what counts as evidence that something is actually real? The advantage of traceability Let’s look at the functions of a news channel. Before a reporter goes live, someone has usually made the phone calls. Someone has spoken to the police officer, ministry official, company spokesperson, lawyer, eyewitness or local correspondent. Someone has looked at the footage. Someone at the desk has asked where the information came from. A producer has decided whether it is strong enough to put on air. An editor ultimately owns that decision. The process may be imperfect, slow. Sometimes it fails miserably too. Television can sensationalise. Newspapers can get stories wrong. Radio can carry propaganda. But there is one important difference. There is a trail. If a report is wrong, there is a publication that carried it, a channel that aired it, a reporter whose work can be questioned and an organisation that can be asked to correct or defend it. Broadcasting standards have explicitly required significant mistakes to be acknowledged and corrected with sufficient prominence. That is very different from a synthetic website whose “reporter” does not exist, whose profile photograph was generated by AI, whose articles are produced at scale and whose owners may disappear the moment anyone starts asking questions. This traceability explains why real audience and real reach is still intact in traditional media. A familiar broadcaster has an audience because people have repeatedly chosen to return to it. The newsroom behind it has accumulated a history alongside that audience. An AI operation can copy the appearance of such an institution quickly but creating the relationship is much harder. The human angle A 2026 study published in the Journal of Media Horizons took 312 digital news consumers and showed them the same news content under different authorship labels. One group was told it was human-produced; the other was told it was AI-generated. The wording, topic, length and factual content were otherwise held constant. The human-labelled news received an average trust score of 4.12, compared with 3.74 for the AI-labelled version. The difference was statistically significant, and the researchers found that authorship explained 18% of the variation in trust. That suggests the journalist's value in the AI age may increasingly lie not simply in writing the paragraph. That AI can do much better. The journalist's value is in being able to explain why the paragraph exists. Where the information came from, who was spoken to, what was checked and why the newsroom decided to publish it. Which traditional media can do much better. Advertisers need to choose wisely This is where the argument becomes commercially interesting. Traditional media may not win by promising more inventory than the internet. Its opportunity may lie somewhere else. If AI makes content abundant and makes some forms of apparent engagement easier to manufacture, then certainty about the environment in which an advertisement appears becomes more valuable. That is a different proposition from the old “TV has reach” argument. Established news brands can potentially offer a bundle of things together. An identifiable publisher, an editorial process, a history with its audience and a distribution system populated by people who were there before the advertiser or today's story turned up. Call it verified attention, accountable reach or simply a real audience with an address. Whatever the industry chooses to call it, it deserves to be priced. An important caveat This is not really a battle between television and the internet. Some of the world's strongest journalism is produced by digital-first organisations. Equally, a reckless television channel does not become trustworthy because it owns cameras and a studio. The deeper divide is between media that can show its workings and content that cannot. Traditional news organisations simply happen to own a lot of the infrastructure that makes those workings visible. Correspondents, news desks, editors, archives, broadcast records and long-standing audience habits. For years, much of that machinery looked expensive compared with the frictionless economics of digital publishing. AI may be changing what that expense represents. The newsroom is not only a content factory. It is also an accountability system. That may be the smarter way for traditional media to think about its role in an AI-saturated world. Not “we are old, therefore trust us.” Not “digital is fake.” The pitch needs to be, “We can show you where this came from. We can tell you who stands behind it. And we have an audience of real people.” Perhaps traditional media spent too long defending its scale when it should have been defending its certainty. Because AI can make more content than any newsroom ever will. What it cannot instantly manufacture is the history between a news institution and people who have chosen, repeatedly, to come back. That may be the real differentiator hiding in plain sight.