How to Protect Your Online Course Videos Without Going Paranoid

The moment you upload your first course, an uncomfortable question appears: what if someone copies it? The honest answer has two parts. First, no video protection is absolute. Second, you can make sharing your course inconvenient, traceable and unrewarding, and that is usually enough. This article walks through the four common layers (signed links, watermarks, DRM and device control), explains what each one does using technical sources, and ends with a proportionate setup for a small course.

What you are actually protecting

Before choosing tools, separate three different risks, because each layer solves a different one:

• The link that circulates. A learner copies the video URL and drops it in a group chat.
• The shared account. A learner lends their login to several people.
• The file copy. Someone records the video and uploads it elsewhere.

The first two can be handled well with ordinary engineering. The third cannot be fully prevented, only made more costly and less attractive. Knowing this keeps you from spending money on solutions designed for film studios.

Layer one: signed, expiring links

A video hosted at a public URL can be watched by anyone who knows the address. The alternative is for the video to require a signature that your server generates only when the learner is entitled to watch.

Cloudflare Stream's documentation puts it plainly: without signed URLs, anyone with the video ID can watch it; with them, viewers need a signed token. The same page says the token returned by its endpoint is valid for one hour by default, that if you sign with your own key the expiry claim cannot be more than 24 hours after signing, and that revoking a signing key invalidates every token created with it.

Amazon CloudFront works on the same idea. A signed URL carries a policy statement with additional information, for example an expiration date and time, and with a custom policy you can also set a start time and restrict access to an IP address or range. CloudFront checks the expiry at the moment of the HTTP request, so a download already under way can finish even if the deadline passes midway, while a new request afterwards fails.

What this gets you:

• A copied link dies on its own after minutes or hours.
• You can cut off access for someone who stops being a learner, without waiting for anything to be deleted.
• Your server decides on every playback whether the learner has a licence, an up-to-date payment and a valid account.

What it does not get you: the signature protects the delivery, not the content once it reaches the player. Anyone authorised during those minutes receives the video and could technically record it while watching.

Layer two: a visible watermark

A visible watermark (the learner's name or email overlaid on the picture) does not stop copying: it identifies. Its value is deterrence and traceability. Someone who knows their name is on screen thinks twice before posting a recording, and if a copy surfaces online you can tell which account it came from and close it.

It matters because it works exactly where the other layers fail. Wikipedia's entry on the analog hole notes that, for digital images, watermarking can be an effective measure.

Good practice for a useful mark that does not ruin the experience:

• Legible but discreet, with moderate opacity, in a spot that does not cover what you are teaching.
• Moving position every so often, so a simple crop cannot remove it.
• Carrying something that really identifies, not just "Confidential": the learner's name or email.
• Drawn in the player or in the video itself, not as a label a technical user can delete with browser tools. That last difference matters: an HTML overlay is easier to remove than a mark embedded in the frames.

If the mark lives in the player, a determined learner can hide it; if it is burned into the video, the processing cost rises. For a small course, an overlay in the player already changes the behaviour of the vast majority of people.

Layer three: DRM

DRM encrypts the video and hands the decryption key only to an authorised player, normally inside a protected module of the device. On the web, the piece that makes this possible is the W3C Encrypted Media Extensions specification, which defines a common API to discover, select and interact with those systems. It helps to read what it does not do: the text itself states that it does not define a content protection or DRM system, and it leaves authentication and authorisation to page authors.

The concrete systems are three:

• Widevine, from Google. Its overview says it follows the EME specification and the common encryption standards, requires a licence agreement and charges no fee for its products and services. It lists Chrome, Firefox, Edge and Android as supported, and desktop Safari and tvOS as not supported. Every licence request must go through a proxy operated by the content provider, and the licence is individualised to the device.
• FairPlay Streaming, from Apple, for HLS on Apple platforms. According to Apple's page, the server SDK requires Apple Developer Program membership, and production credentials are approved only if the team provides a streaming service to consumers; requests from third parties acting on behalf of content owners are not approved.
• PlayReady, from Microsoft. I could not check it against an official page, so I give no details here.

The practical consequence is that full DRM across every browser and device means combining several systems, a licence infrastructure and contractual conditions. That is why it is usually bought as a managed service, and why it is sensible to ask whether your course justifies it.

Strength also depends on the device. The Android documentation explains that the DRM plugin can use hardware-backed protection, but that security relies on hardware, software and services working together, and not every platform supports every feature. That is why many services cap resolution on devices with fewer guarantees.

The real limits: screen, HDCP and the analog hole

This is the part protection vendors tend to tell halfway.

Screen recording and cameras. DRM stops the encrypted file being extracted, but at some point the video is displayed. The definition of the analog hole sums it up: it is a perceived fundamental and inevitable vulnerability, because once content is made visible or audible, that output can be digitally recaptured without the protections of the original. A screenshot or a photo of the screen is always possible, even if quality drops.

HDCP. It protects the cable between devices, not the screen. According to the Wikipedia entry on HDCP, it encrypts the signal between a transmitter and an authorised receiver on interfaces such as HDMI or DisplayPort. The same source records that a master key was published in 2010, which made key revocation useless, that flaws have been demonstrated in several versions, and that devices exist that remove HDCP protection from the signal.

Bottom line: an attacker with time and skill can copy almost any video that can be watched. What protection changes is the cost, the risk and the traceability of doing so.

What proportion makes sense for a small course

Think about who is likely to copy you. In a movement, medical or coaching course, the common scenario is not a piracy ring but a learner who shares their account or forwards a lesson to a colleague. The proportionate defence for that is:

1. Signed links with a short expiry. Cheap, nearly invisible to the learner, and it removes the circulating link. Make the expiry cover the lesson length plus a reasonable margin.
2. A watermark with the learner's identity. The most effective measure against screen recording, because it turns an anonymous copy into a traceable one.
3. A limit on devices and simultaneous playback. It curbs account sharing, which in practice is the most common leak.
4. Clear terms of use. A personal licence in writing, accepted at purchase, gives you a basis to act if you detect a leak.
5. Value that cannot be copied. Personal feedback, community, one-to-one sessions and updates: none of that fits in a video file.

Full DRM makes sense when the video is your only product, has a high price per unit and there is a credible risk of organised resale. If your course does not meet those three conditions, it adds complexity, playback errors on older devices and cost, in exchange for a benefit the analog hole limits anyway.

One last criterion: measure the possible damage before choosing the defence. If a leak would cost you a few sales, it is not worth making the experience worse for paying learners to stop someone who was never going to pay.

What LearnSuite does

In LearnSuite, paid lessons are served through short-lived signed links, so a copied URL expires. You can turn on, per course, a watermark with the learner's details over the video, and each licence allows up to three devices and one playback at a time. We do not present this as an impassable barrier: it is a proportionate combination that makes account sharing harder and leaves a trail if someone records.