In this article
Step 062 covered the switch itself: what restricting saved content removes, what it costs the members who were never going to leak anything, and the limit that matters — the protection is real inside Telegram and absent outside it. A second phone pointed at a screen defeats all of it, and always will.
This step starts where that one ends. If the copy cannot be prevented, the remaining questions are different ones: can a copy be traced back to whoever made it, what does that actually buy you, and what do you do on the morning you find your material somewhere else?
What you will be able to do after this step
- check_circle Tell prevention and attribution apart, and choose one deliberately.
- check_circle Say what a per-member watermark forces you to change about delivery.
- check_circle Mark material in a way that survives a crop and does not falsify it.
- check_circle Act on a leak without expecting the platform to fix it.
security Prevention and attribution are different projects
Prevention stops the action. That is the setting from Step 062, and it works in exactly one place: the inside of the Telegram application, where the client software is obeying a flag on your channel. Nothing beyond that boundary is under its authority.
Attribution makes the action identify the person. A visible mark carrying somebody's name travels with the picture everywhere the picture goes, including into a photograph taken of a screen with another phone. It cannot stop anything at all — it changes who is willing.
That distinction is the whole step, because the two answer different worries. The member who would casually pass a chart to a colleague stops when their own name is written across it. The person who intends to redistribute your work publicly is entirely undeterred, crops what they can and posts the rest. So before choosing a tool, name the behaviour you are actually worried about, because attribution is enormously effective against the first and worth almost nothing against the second.
And be clear about what attribution is not. It is not a lock, not a licence and not enforcement. It gives you a name and leaves you holding it.
perm_media A per-person watermark changes how you deliver
This is the consequence nobody mentions, and it usually ends the idea. A mark that identifies a person has to be different for every person. A channel post is one file shown to everybody. Therefore a per-member watermark cannot be a channel post at all.
Which means the material has to be generated and delivered individually — on request, through a bot, one copy per member. You have swapped a channel for a small distribution service, complete with a queue, storage, failed deliveries and people writing to say theirs never arrived. For a handful of expensive documents that is a reasonable trade. For anything you publish daily it is an absurd one, and the people who try it usually stop within a month.
If you do mark per person, put it where a crop cannot reach: across the middle of the thing people want to look at, not tucked into a corner. A corner mark is removed in two seconds by anybody with a phone, which makes it a mark that only inconveniences honest readers — the worst of both projects.
Your channel has saving restricted. A member shows you a photograph of one of your posts, taken on a friend's phone. What failed?
The setting is enforced by the Telegram client, on the device, for actions the client controls: forwarding, copying, saving, and screen capture where the operating system offers a way to block it. A camera pointed at a screen involves none of those. This is not a bug, a version problem or a platform gap — it is the outer edge of what any software protection can reach, and it holds for every service that claims otherwise. Build your expectations on it rather than on the marketing, because a plan that assumes the material cannot leave the screen has a hole in it the size of every phone in the room.
You write each subscriber's name across your report. One of them redistributes it anyway. What has the watermark bought you?
You learn who. You do not get the file back, you cannot make anyone delete their copy, and you have no way of knowing who they sent it to. What you can do is remove that person from your channel, which is worth doing and is not a remedy. That is the honest value proposition of forensic marking: it is a deterrent that works before the fact on people who are weighing it up, and a name on a piece of paper after the fact. Anybody selling it to you as protection is selling the wrong noun.
You want free posts forwardable and paid posts locked, in the same channel. Possible?
The switch in the channel's settings is all-or-nothing and applies to everything already published as well as everything to come. Protection on individual messages exists, but it is a property set when the message is sent, which in practice means the message has to be published by a bot with posting rights — the same arrangement Step 081 describes for buttons. That is the configuration most publishers actually want: the public posts stay forwardable, because forwarding is how a channel grows, and only the material somebody paid for is locked. It is also a reason to set up a bot that is nothing to do with buttons.
travel_explore Finding a leak, and what you can actually do
Public redistribution is visible by definition — somebody has to publish it somewhere findable for it to hurt you. Searching for a distinctive phrase from your own material, occasionally, finds most of what there is to find.
A quieter method is to vary something harmless between copies so that a recovered copy identifies its source without carrying a visible mark. Vary the cosmetic, never the substance. Different wording of the same sentence, a different ordering of two equivalent items, a different filename. What you must not do is plant a false figure, a wrong date or a fabricated detail: the moment you do, you have published something untrue to catch somebody, and the person it lands on is a reader who believed you. That trade is never worth making, and it is exactly what the "unique price target" version of this advice proposes.
When you find a copy, the options are short and honest. Inside Telegram, you can report a channel republishing your work, and you can remove and ban whoever leaked it — and unlike the situation in Step 078, this is the case where the ban should stay. Outside it, you are talking to whoever hosts the file. Neither route is fast, and neither is certain.
What no route offers: retrieving the file, obliging anybody to delete their copy, or discovering who else received it.
touch_app Try it now — five minutes
Four steps, and the first one settles the argument faster than any amount of reading.
- Photograph your own protected post with a second phone Ten seconds. That image is your threat model, and everything you decide afterwards should assume it exists.
- Name which project you are doing Preventing casual copying, or being able to identify a copy. They lead to different tools, and trying to do both badly is the usual outcome.
- Put your channel address on your images Small, legible, and positioned where removing it would take content with it. This is the highest-value five minutes on the page for most publishers.
- Search for one distinctive sentence of your own You will either find nothing, which is useful, or find something, which is more useful. Either way you now know rather than assume.
verified The protection that actually holds
Three properties make material resistant to leaking, and none of them is a setting.
Value that decays. Something that matters this morning is worth nothing by Friday. A leaked copy of a time-sensitive piece is a leak of an expired thing, which is why publications built on being current have a leaking problem that solves itself.
Value that is access rather than a file. A course can be copied once and passed around forever. A course plus a room where you answer questions cannot, because the room is the part people are paying for and it does not copy. If your material is inherently copyable, the sensible move is to attach something that is not.
A price people think is fair. The population willing to redistribute overlaps heavily with the population that felt overcharged. A modest price with no lock leaks less than a steep price with a heavy one, and the steep-and-locked version also loses you the honest readers who found the restrictions insulting.
Which returns to the arithmetic from Step 062 and settles it: the restriction is paid for by every honest member, every day, and the leak costs you once. Most channels that switch it on switch it off again within a few months, and they are not being careless. They are counting correctly.
help Common questions at this stage
Does screenshot blocking work on every device?
No, and this varies by platform in ways you cannot control or detect from your side. Some systems provide a way for an application to mark a window as protected, and some do not; some notify the person afterwards instead. Treat screenshot behaviour as unreliable everywhere and plan as though every post can be captured, because on some device belonging to some member, it can.
Is a watermark worth it for images I post daily?
A channel mark, yes — it costs nothing per post and turns circulation into advertising. A per-person mark, no. Daily material cannot be delivered individually without abandoning the channel format entirely, and the value of any single daily post is too low to justify running a delivery service around it. Save individual marking for the rare, expensive document.
Can I stop people taking my text and reposting it as their own?
Not technically. Text is the easiest thing in the world to retype, and restricting copying only means somebody types it instead of pasting it. What works better is being identifiable in the writing itself — a voice, recurring framing, references back to your own earlier posts — so that a stripped copy reads as second-hand. Readers notice this more than publishers expect.
Should I convert documents to images so they cannot be edited?
It stops editing and copying of text, and it costs you a great deal: images cannot be searched, cannot be read aloud by a screen reader, cannot be translated by the tools in Step 053, and are painful to read on a phone. You are imposing a permanent accessibility penalty on every legitimate reader to slow down a copy that a screenshot would produce anyway. For most material this is a bad trade, and for material that has to reach people with visual impairments it is not a trade at all.
Somebody is reposting my channel wholesale. What is the realistic outcome?
Report it, and expect a slow and uncertain process rather than a switch. In the meantime, the practical response is the one that does not depend on anybody else: publish where the copy cannot follow. A reposting channel republishes your posts; it cannot republish a live session, a discussion room, a reply to a question, or anything that happens rather than exists. Channels that get copied and keep growing do so by being somewhere rather than by being something.
Does any of this stop a member screen-recording a live broadcast?
No more than it stops a photograph, and less than you would hope, because a live session is long and a recording of it is valuable to somebody. The mitigations are all social: say at the start that the session is not to be recorded, keep the genuinely sensitive part for people you know, and recognise that the things worth attending live — questions answered in the moment, a conversation — are the parts a recording reproduces worst. That is the same principle as the section above, applied to Step 079.
At a glance
- 1Prevent or attribute
Content protection blocks saving; a watermark only names the source.
- 2Mark the middle
Put a mark where a crop cannot remove it.
- 3Do not falsify
Change appearance to trace a leak, never the facts.
Updated
