YouTube Watch With: Another Creator Can React Live to Your Video and Keep the Ad Revenue

For twenty years, the way to make money from someone else's YouTube video was to download it, talk over it, upload the result and hope. Hope the Content ID match did not land. Hope that if it did, the claim took the ad revenue rather than the video. Hope the rights holder was the kind who tolerated commentary. That gamble built an entire genre, and nothing about it was ever settled.

On 23 September 2026, YouTube announced a different arrangement. Watch With, the co-viewing feature it had been piloting around sports and then testing with mobile creators, is being extended to eligible long-form videos across the platform. A creator will be able to open your video, tap "React live", and broadcast their own stream with your video playing inside it. They keep the ad revenue from that broadcast. You get credit.

That is a bigger change than the one-line summary it received in most of the coverage, including in our own write-up of the announcement, which had three dozen features to get through. This post is the long version: what Watch With actually is, who earns what, the default setting that decides most of the outcome, what "view credit" would have to mean to be worth anything, how it compares with the way co-streaming has worked everywhere else, and whether you should leave it switched on.

What Watch With is, and where it came from

Watch With did not start as a creator tool. It started as sports programming. When YouTube streamed its first exclusive free global NFL game — Chiefs against Chargers in São Paulo, on 5 September 2025 — it ran creator-led companion broadcasts alongside the main feed. Reporting at the time named iShowSpeed and Tom Grossi commentating in English, Robegrill and SKabeche in Spanish, with CazéTV running its own Portuguese broadcast from the stadium.

The logic there is old television logic: one event, several commentary tracks, pick the voice you like. What was new was that the voices were creators with their own audiences, and the companion streams were their channels rather than a network's second feed.

The feature then moved from events to ordinary streaming as a mobile experiment. Creators in the test group saw a "React live" button in the share panel while watching an eligible livestream on a phone. Tapping it started their own broadcast with the source stream composed into it. Reporting on that test described some specific constraints: the phone camera locked to a horizontal orientation, the camera could not be switched off mid-reaction, and viewers saw the combined result as a vertical stream whatever device they watched on. YouTube did not publish eligibility criteria for the test, and still has not.

So the announcement at Made on YouTube was the third step rather than the first. The pilot proved co-viewing worked for a scheduled event. The mobile test proved it worked for arbitrary livestreams. The expansion applies it to the thing YouTube has in effectively unlimited supply: long-form video on demand.

What was announced, and when each piece lands

Three live features came out of the event, and they are easy to confuse because all three involve two creators and a shared audience.

Feature	What it does	Timing as announced
Watch With expansion	React live to eligible long-form videos, not just livestreams	Rollout starting later in 2026
Go Live Together matchmaking	Browse creators streaming right now, or join a queue to be paired automatically	Early 2027
Live showdowns	Two co-hosting creators compete; chats, gifts and Super Chats drive an on-screen score	2027

Only the first is imminent, and only the first involves someone else's finished video. Matchmaking and showdowns are both consensual by construction: two creators agree to share a broadcast. Watch With in its new form does not require the source creator to agree to anything, which is why it is the one worth understanding properly.

YouTube's own announcement of the live tools frames all three as growth for live. The trade press read the Watch With half differently. PPC Land's headline was that live reactors will keep ad revenue from other people's videos, and Tubefilter described it as YouTube letting creators make reaction content on-platform using everyone else's long-form videos. Both are accurate descriptions of the same feature.

Who earns what

The economics are the part to get exactly right, because the summary "reactions are monetised" hides which side the money lands on.

	Reacting creator	Source creator
Live ad revenue from the reaction	Keeps it	None stated
Fan funding during the reaction	Keeps it	None stated
Views and watch time	Counts on their own stream	Credited back, per YouTube's description
Attribution on screen	Is the host	"Clear attribution" to the source
Control	Chooses what to react to, from what is eligible	Controls which of their videos can be reacted to

Read the first two rows together with the third and the shape becomes clear. The reaction is a separate broadcast with its own monetisation, sold against the reactor's audience. The source creator's compensation is measurement: concurrent viewers, public views and watch time from the reaction are reported as crediting the original content, and the source creator sees the effect in Studio.

That is not nothing. A view is the input to most of what matters on YouTube. But it is also not revenue, and YouTube has not said whether credited views behave like ordinary views in the places that pay — whether they carry ad impressions, whether they count towards Partner Programme thresholds, whether they feed the recommendation systems that decide what your video does next week. Those are three different questions and the announcement answered none of them.

The default is on, and that decides most of this

In the mobile test, eligible streams allowed reactions by default. A creator who wanted none had to change a setting from "Allow eligible channels" to "No one", on mobile, before going live. Nothing about that is unusual for a platform experiment, and everything about it matters, because defaults are the policy. The share of creators who will read a Studio changelog, find the toggle and make a considered decision is small. The share who will discover the feature when a notification tells them someone is reacting is much larger.

Whether the long-form expansion inherits the same default is the single most consequential unanswered question about it, and YouTube has not said. The announcement's language — that source creators get control over which of their videos can be used — tells you a control exists. It does not tell you which way it points when you have never touched it.

If you stream, check the setting before your next broadcast rather than after. If you publish long-form and the expansion reaches your account, expect the control to live in the same place as your other per-video permissions, alongside embedding and the other things nobody looks at until they matter. Our guide to going live on YouTube covers where the stream-level settings sit today.

What "view credit" would have to mean to be worth anything

There is a useful precedent for how slippery combined measurement gets, and it comes from Watch With's own launch event.

YouTube said the NFL Brazil game averaged 17.3 million viewers, a figure built from internal data plus Nielsen measurement. Analytics executives at rival networks publicly objected that Nielsen had built YouTube a bespoke measurement product. Then the number moved: as The Desk reported, YouTube acknowledged it had under-reported the audience to Nielsen because of a technical fault, and the figure was revised upward to roughly 19.7 million. Reporting at the time also raised the question of whether the headline figure included the creator companion streams at all.

That episode is not a scandal so much as an illustration. The moment one audience is watching the same content through several channels, "how many people watched it" becomes a definitional choice rather than a count. Watch With at scale asks every creator to accept the same kind of definitional choice about their own analytics, made by the platform, with the method unpublished.

So a short list of what would make credited views genuinely valuable, in rough order of how much it would matter:

1. Monetised. Credited views carry ad impressions that pay the source creator, at any rate above zero.
2. Algorithmically live. Credited watch time is an input to recommendations for the source video, so a reaction can push the original back into circulation.
3. Separable in Studio. A traffic source row that says how much of this came from reactions, and from whose, so the effect can be measured rather than assumed.
4. Counted for thresholds. Credited hours count towards the Partner Programme watch-hour requirement.
5. Attributed with a link. Attribution that a viewer can actually click, so some fraction of the reaction's audience lands on your channel.

Note that the fifth is the one creators instinctively care about and the one that depends least on YouTube's accounting. If the reaction sends traffic, you will see it in your own traffic sources whatever the credit scheme does.

This is a permission change, not a copyright change

It is tempting to read Watch With as YouTube settling the reaction-video question. It does not. Copyright law is exactly where it was. What has changed is that there is now a sanctioned route that never engages the question.

The legal test for reaction content has always been transformation: whether the new work adds commentary, criticism or meaning rather than repackaging the original. A 2024 analysis in the Fordham Intellectual Property, Media & Entertainment Law Journal works through how that test applies to reaction content specifically; the practical reading creators arrive at is that a reaction which pauses, critiques and analyses stands on much firmer ground than one that plays the source through while someone nods along. YouTube's own copyright documentation is blunt that automated systems like Content ID cannot decide fair use, which is why claims land on videos that would win the argument if an argument were ever held.

Watch With sidesteps all of that by making the source creator's permission a platform setting instead of a legal question. Inside the feature, the reactor is not asserting a defence; they are using an authorised pipe. Outside it, nothing moved: a downloaded clip in an edited upload is still governed by the same claims and disputes described in our guide to copyright claims versus strikes, and a wholesale re-upload is still the problem described in what to do when someone re-uploads your video.

The practical consequence is a two-tier reaction economy. Reactions made inside Watch With are clean, live, monetised for the reactor and credited to the source. Reactions made the old way are unchanged, with all the old risk. If you are a reactor, the incentive to work inside the feature is obvious. If you are a source creator, the thing to notice is that your copyright position is no weaker than before — but your default consent may now be yes.

How this differs from co-streaming everywhere else

Co-streaming is not new. Esports has run it formally for years, and the shape of those programmes is instructive, because YouTube's version inverts nearly every parameter.

	Esports co-streaming	YouTube Watch With
Permission	Applied for, granted per event	A setting on the source content
Scope	One tournament, fixed dates	Eligible videos and streams across the platform
Who may co-stream	An approved list	Eligible channels, criteria unpublished
Source feed	Supplied by the rights holder	The video itself, composed by YouTube
Conduct rules	Explicit: stay on the broadcast, no disparagement	Ordinary platform policy
Measurement	Aggregated for the main broadcaster	Credited to the source, method unstated

Twitch's own documentation of co-streaming describes that world: rights holders approve participants, the main broadcaster sees aggregated viewership across approved co-streams, and the rules of engagement are written down. ESL requires an emailed request naming the event and the channel, restricts co-streams to Twitch, and requires the official broadcast feed rather than a grab from another platform's player. Riot has run its own watch-party programme with an exclusive list.

Those programmes are built around scarcity: a few rights holders, a few hundred approved partners, one event at a time. YouTube is building the same mechanic as infrastructure, for a library in the billions, with the permission question compressed into a toggle. That is the genuinely novel thing here, and it is why the default matters more than any of the revenue details.

What you need to be the reactor

If the appeal of this is on the reacting side, the entry requirements are mostly the ordinary live ones, which are stricter than people expect.

Reacting in the mobile test happened from a phone, and mobile live streaming on YouTube requires at least 50 subscribers for creators aged 18 and over — 1,000 for 13 to 17 year olds — a gate that does not apply to streaming from a computer. Keeping live ad revenue requires being in the Partner Programme at all, so the thresholds in our guide to the 2027 monetisation requirements are the real floor for the money half of this. YouTube has not published what makes a channel "eligible" to react beyond that.

Then there are the production constraints, which are unusually prescriptive for YouTube. A locked camera orientation and a camera that cannot be switched off mean the reaction is a face-on-camera format by design. There is no version of this where you narrate invisibly over someone's video; your face is the point, and it is in frame for the duration. Creators whose whole proposition is not being on camera should read that as the feature not being for them.

What it does to packaging

A reaction stream is a second listing. It has its own title, its own thumbnail, its own place in the feed, and it competes for the same impressions as the video it is reacting to. For the reactor, that is an opportunity: the audience searching for a hot video is large, and arriving with a live take while the conversation is fresh is a legitimate way to be found. For the source creator, it means your topic now has more doors, and you own only one of them.

Three consequences follow for anyone making thumbnails.

The reactor's thumbnail has to signal "reaction" and "subject" in one frame. That is a two-object composition — the thing being reacted to, plus the reacting face — at the size where only one object usually survives. It is the hardest layout in thumbnail design, and the common failure is a split-screen that reduces both halves to mush at feed size. The split works when one half carries recognisable shape and the other carries recognisable expression; it fails when both halves are busy.

Live changes what the thumbnail is for. A live listing competes with a red badge attached to it and a viewer count that does some of the persuasion for you, and the cover you want during the broadcast is often not the cover you want on the replay. We worked through that separately in the guide to live stream thumbnails, and all of it applies to a reaction stream, with the extra wrinkle that your subject matter has an expiry date.

Your thumbnail may now be someone else's set dressing. If reactions compose your video into another broadcast, your packaging appears inside their frame, at a size you did not design for, next to a face you did not choose. There is nothing to do about that except what you were already doing: make the thumbnail legible small and recognisably yours. A composition that reads at 120 pixels wide survives being cropped into somebody else's layout; one that depends on four lines of text does not. If that is the part of your workflow that is weakest, generating and comparing options against each other is the fastest fix, which is what Thumblore exists to do.

Should you leave reactions on?

There is no universal answer, and anyone offering one is not accounting for what different libraries are worth. The question to ask is whether a live reaction to your video substitutes for watching it or advertises it.

• Tutorials, explainers and reference content: leave it on. A reaction cannot substitute for the thing someone came to learn, and anything that puts your video in front of a new audience mid-explanation tends to send that audience to the source.
• Narrative, documentary and anything with a reveal: be careful. If the value of the video is the order in which information arrives, a reaction stream delivers the payoff with someone else's face over it, and the viewer has no reason to come back for the original.
• Performance, music and comedy: be careful for a different reason. These are formats where the reaction genre has always generated the most friction, and where a live co-view is closest to a straight substitute for the work.
• News, commentary and anything time-sensitive: leave it on. Your window is hours, reach matters more than exclusivity, and a reactor pulling their audience into your story while it is live is a net gain.
• Licensed or commissioned footage: switch it off until you have checked. If your video contains material you licensed, or a sponsor's assets, you may not have the right to authorise a third party to broadcast it inside their monetised stream. That is a contract question, not a YouTube question.

Notice that this is not a judgement about whether reaction content is good. It is a question about substitution, and the answer varies by format in ways the single platform-wide toggle cannot capture.

What is still unknown

Writing honestly about a feature that has not shipped means being specific about the gaps. As of early October 2026, YouTube has not said: whether long-form reactions are on by default; how view credit is calculated or whether it monetises; what makes a channel eligible to react; whether reactions can be limited per video rather than per channel; whether the reaction's archive stays up and who owns it; whether Shorts are in scope; or whether source creators get a notification when a reaction starts.

Any of those could reasonably change the right decision. A notification plus a per-video control is a feature most creators would accept. No notification, a channel-wide toggle and unmonetised view credit is a different product with the same name.

What to do this week

Three things, in order. Check your live stream settings and set the reaction permission deliberately rather than by default, so that whatever the long-form version inherits, you have made the choice once. Decide which of the five categories above your library falls into, because the decision is much easier before someone is already live with your video. And if you are tempted by the reactor side, be honest about whether a face-on-camera, fixed-orientation, time-sensitive format is one you can actually sustain, or whether it is simply the newest thing.

The deeper point is the one the announcement did not make. YouTube spent 2026 building tools that decide things on creators' behalf — thumbnails generated from a transcript, cuts tested on live audiences, images matched to audience segments, and now permission to use your work set by a default you did not choose. Each one is individually defensible and the direction is unmistakable. The response is not to refuse the tools; it is to know which switches exist and to set them yourself.

If this is the first you have heard of the wider announcement, the full list is in our Made on YouTube 2026 breakdown, and the packaging half of live is covered in the live stream thumbnails guide. Whichever side of a reaction you end up on, the thing that decides whether the stream gets found is still the frame and the sentence attached to it.