The Laura Rutledge AI Video, Explained — and Why Her Response Worked

September 24, 2026 · 6 min read

The Laura Rutledge AI video is an AI-altered clip of a real postgame interview, edited so that Chargers quarterback Justin Herbert appears to punch ESPN sideline reporter Laura Rutledge to the ground — an incident that never happened. The fake first circulated after a genuinely awkward Monday Night Football interview in December 2025, and it is trending again in September 2026 because commentators keep contrasting Rutledge's pitch-perfect reaction with the fallout from the Arch Manning–Holly Rowe AI video. Here is what actually happened, why the fake convinced people, and what her response got right.

What actually happened

The real event was uncomfortable but entirely verbal. In December 2025, after the Chargers beat the Eagles 22–19 in overtime on Monday Night Football, Rutledge chased down Justin Herbert for a postgame interview. Herbert was curt, gave clipped answers and walked off — and the genuine clip of the exchange went viral on its own, with Herbert widely criticised as rude (People, December 9, 2025).

Someone then took that real broadcast footage and altered it with AI: in the fabricated version, as Rutledge asks "Can we just talk really fast?", Herbert appears to punch and slap her to the ground. None of that occurred. No such contact happened, no outlet that aired the game reported it, and the only footage of it is the edited clip.

Rutledge's response is the reason the story has a second life. Rather than demanding takedowns or issuing a statement through ESPN, she retweeted the fake herself and posted on X: "Rough day at the office 😂 let's all have a good laugh and move on!" The post was widely praised as the model response — unbothered, funny, and impossible to scandalise further because she had already owned the joke.

Why is a December 2025 fake trending in September 2026? Because of the Sarkisian–Holly Rowe AI video and Arch Manning's apology for laughing about it (covered in our explainer). Since September 19, commentators — including Sage Steele on her podcast — have held up Rutledge's handling of her own AI fake as the contrast case: the target who laughed it off versus the incident that produced criticism, an apology and a statement from a players' union of reporters (EssentiallySports via Yahoo, September 20, 2026). That discourse pushed "laura rutledge" back up the trends, and the old fake is circulating again with it.

Why these fakes convince people

The Rutledge–Herbert clip works for the same reasons the Sarkisian–Rowe one did, and they are worth naming because this is now a repeatable genre, not a one-off:

• It edits real footage rather than generating a scene. The stadium, the lighting, the camera work, the faces and most of the audio are genuine broadcast video. Only a few seconds of action were changed, so the visual "tells" people are trained to look for barely exist.
• It amplifies something viewers already half-believed. The real interview had aired live and Herbert had been criticised for it. The fake doesn't ask you to accept a new event — it offers a worse version of one you already remember being awkward.
• It is watched under the worst possible conditions: short, loopable, re-uploaded, compressed to mush on a phone screen — exactly where remaining artifacts disappear.

How a clip like this is made

The short version, at the level of understanding rather than instruction: this is an AI-assisted edit of real footage, not a full deepfake. The editor takes the genuine broadcast clip, masks the region where the interaction happens, and has a modern video model regenerate just those seconds with a different action — the models hold identity, clothing, lighting and camera motion consistent across the edit, because temporal consistency is what they are trained for. The real broadcast audio stays, the clip is compressed and re-uploaded, and the seams smear away. Our Arch Manning explainer walks through the technique in more detail, with a checklist for spotting it.

How to check this specific genre of fake

This category has one decisive weakness: the original exists and aired live. Before anything else:

1. Find the broadcaster's footage. The real Herbert–Rutledge interview aired on Monday Night Football and was replayed everywhere — anyone who pulled it up could see there was no contact. An edit of real footage can never survive comparison with its source.
2. Check what the people involved said. Rutledge commented publicly on the fake itself — the person allegedly assaulted was laughing about it on X. The subject of a video like this is the fastest credible source there is.
3. Watch the contact physics. Does her balance, hair and clothing react at the exact instant of the supposed hit? Fakes routinely land a punch with no recoil, or start the fall a beat early.
4. Ask who posted it first. A clip of a nationally televised assault that exists only as anonymous re-uploads — never from ESPN, the NFL, or anyone at the stadium — is unverified by default.

The line this crossed: consent

It is worth being direct about this, because we make AI video tools. Hermosso exists to transform your own photos and videos, with your consent: you upload your selfies, train a private model of your face, animate your own pictures. Fabricating violence against a real, identifiable person without their consent is exactly what this product is not for — and in a growing number of jurisdictions it is explicitly illegal, regardless of whether the target laughs it off. Rutledge's grace made her story end well; it does not retroactively make the fake harmless, and the next target may not want the joke.

If what you actually want is the legitimate version — putting yourself in a clip — Photo to Video animates your own photo into a short clip for 250–2,400 credits per 5 seconds depending on the engine, and our talking-photo guide covers turning a picture of yourself into a speaking video. New accounts get 400 free credits, and plans start at $9/month for 5,000 credits — the full breakdown is on the pricing page.