We laughed at China’s robotic news reader in 2018. Eight years on, India’s most decorated anchor doesn’t exist, Mexico runs a stable of synthetic presenters with fan bases, and I’m left wondering whether we automated the wrong half of the job.

By Ben Anchor — Wednesday, 8 July 2026 · Listen to the podcast episode

There’s a moment in every technology panic where the industry stops laughing and starts calculating redundancy costs. For synthetic presenters, that moment was somewhere between 2018’s stiff Chinese news avatar and 2026’s Sana — India Today’s AI anchor, hundreds of broadcast hours deep, multiple languages, primetime slot, actual awards on the mantelpiece. She doesn’t exist, and she’s better at parts of my job than I am.

I recorded an episode this week on AI anchors, and three things have stayed with me since. Not the obvious stuff — yes, the technology works now; yes, it’s cheaper; yes, someone’s job is disappearing. The three things that won’t let go are harder than that. They’re about what we thought an anchor was for, what the audience was actually trusting, and whether we’ve just spent seventy years confusing the two.

We Automated Everything Around the Presenter Until There Was Nothing Left to Automate But the Presenter

The first thread is a confession. Years ago, I stood on a studio floor and watched pedestal cameras gliding around on their own. No operators. Every move automated, night after night, executed to the frame. I thought it was revolutionary. It was. But here’s what it actually was: the industry methodically automating everything around the anchor until the anchor was the last human left on the floor.

The synthetic presenter is that same logic finally reaching the desk itself.

And the business case is so obvious it barely needs stating. Scale, speed, consistency. Grupo Fórmula in Mexico built Nat for entertainment, Sophie for politics, Max for sport. Each avatar fronts five or six stories a day across vertical video platforms. Sophie’s had videos clear a million views. That’s a volume of on-camera output no human roster could sustain at that cost. The machine doesn’t need make-up, doesn’t get flu, and it publishes at the speed of the story rather than the speed of the studio booking.

But here’s where I part company with the efficiency argument. Because the Mexican team says the same thing in every interview: five humans sit behind Nat doing the writing, the checking, the post-production. Five people to run one fake person. So this isn’t a zero-staff operation. It’s a zero-presenter operation. The job didn’t vanish — it moved off-camera. And that distinction matters, because it tells you what we actually automated: not the journalism, just the face.

The Anchor Was Never One Job — We Just Pretended It Was

The second thread is about trust, and it’s the one that makes people angriest. News is a trust business. The entire point of an anchor — the reason the word exists — is that a known human being looks down the lens and stakes their reputation on what they’re about to tell you. How does an audience trust a face that can’t be embarrassed, can’t resign, and can’t be wrong in any way that costs it anything?

The uncomfortable evidence: ask audiences how they feel about AI in news, and the pattern is consistent. People are relaxed about AI in the plumbing — transcription, translation, tagging. Their comfort collapses the closer the machine gets to the front of the camera. The face reading the news is the last place people want the synthetic to sit.

And I think the audiences are right, but not for the reason they think.

Because the anchor job was never one thing. Part of it is transmission: reading accurate words clearly, around the clock, in as many languages as your audience speaks. The machine is already better at that part. But part of it is witness: judgement in the moment, the raised eyebrow at the evasive minister, the authority to say “we don’t know yet” and be believed. That part doesn’t automate, because its entire value is that a human is spending their reputation in front of you.

The broadcasters getting this right are unbundling those two jobs. Let the synthetic layer carry the transmission. Spend the humans where witness matters — the interviews, the investigations, the terrible nights. Nobody wants the synthetic face on the night the towers fall or the king dies, and nobody serious is proposing it.

The mistake isn’t using the machine. The mistake is using it where witness was the product.

We’re About to Graduate a Generation Who’ll Never Get the Job the Machine Is Already Doing

The third thread is the one that kept me awake. There’s a reported piece out of China that won’t let go: broadcast schools there are still full of students training in the classical anchor skills — breathing, diction, composure. They’re graduating into a market where regional stations have already licensed digital clones of their star presenters. A famous host sits for a scanning session. The station gets a version of them that works around the clock. And the next generation is left hustling for a job the copy is already doing.

The entry-level rungs of the presenting ladder are exactly the rungs the machine eats first. You don’t lose the star. You lose the path that made the star.

And this is where the industry advice gets slippery. “The job is moving, not vanishing,” we say. “The ones who thrive will understand the machine well enough to stand next to it.” All true. But it’s also true that we’re about to discover we built an entire training pipeline for a job that no longer exists at the bottom.

The delivery skills alone won’t carry a career anymore, because delivery is exactly what got automated. The premium now is on being verifiably, accountably human — the judgement, the reporting, the ability to hold a live moment and be trusted in it. Those are scarcer than ever.

But let’s not dress it up: we’re not creating more jobs for that scarce human talent. We’re creating fewer, harder-to-reach jobs, and telling everyone it’s a meritocracy.

So What’s an Anchor Actually For?

After all this, here’s where I land. An anchor is not the thing that moves. It’s the thing that holds you steady while everything else does. For seventy years, we bundled that steadiness into a person behind a desk. We’re now discovering how much of the bundle was ritual and how much was real.

The reading can be synthesised. The steadiness can’t, because steadiness is a promise, and only someone with something to lose can make a promise.

The broadcasters getting this right are doing three things:

1. Writing disclosure policies before commissioning the technology. How is a synthetic presenter labelled on screen, in the feed, in the metadata? Treat next month’s European transparency rules as the floor, not the ceiling.

1. Drawing the line between transmission and witness in writing. Make it an editorial policy, not a budget accident made at midnight. Decide now what a synthetic face may never front.

1. Getting likeness rights into contracts today. Who owns a presenter’s face and voice? What’s a scanning session worth? What happens to the clone when the human leaves? Your star’s digital double is a rights negotiation waiting to happen, and it’s cheaper to have it now than in a courtroom later.

I’m not frightened of the AI anchor. I’m frightened of broadcasters forgetting which part of the job the audience was actually trusting. Get that right — label the machines, spend the humans on the moments that matter — and there’s room on the desk for both.

Which, given my surname, is a sentence I never expected to say out loud.