Short version up front: downloading a caption or transcript for yourself — to take notes, search a long video for the part you need, or make content more accessible — is broadly low-risk and something millions of people do every day. Republishing that transcript, word-for-word, somewhere public without permission is a different situation, and the legal analysis gets genuinely more complicated. This page walks through why, and where the real lines sit.
This is general information, not legal advice. Copyright law varies by country, fair use and fair dealing are fact-specific tests applied case by case, and platform Terms of Service change without much warning. If you have a specific situation with money or a business relationship riding on it, talk to a lawyer who can look at your actual facts — not a page like this one.
Who actually owns the copyright
A caption or transcript is not a separate piece of intellectual property that belongs to nobody. It's a text version of something someone already said on camera, and copyright in the underlying spoken content generally sits with whoever created it — the video's creator, the speaker, or whoever they've assigned rights to. Not the platform hosting the video, and not a tool that extracts the text.
That matters because a transcript is legally treated as a derivative of the original work. Extracting the words someone spoke and turning them into a text file doesn't create new rights for you, and it doesn't strip the original creator's rights away either. If the video is copyrighted — and almost all original video content is, automatically, the moment it's created, with no registration required in most countries — the transcript inherits that status.
There's a genuinely interesting wrinkle here worth knowing about: a lot of captions are auto-generated by the platform's own speech-to-text system, not typed by a human. Purely machine-generated text with no human creative input is a harder case for independent copyright protection in some jurisdictions (courts have wrestled with whether AI-generated output can be copyrighted at all). But that's mostly academic for this question — even if the caption file itself sits in a legal gray zone, it's still a transcription of the speech, and the speech is the copyrighted work. Whether the caption text was typed by a person or generated by an algorithm doesn't change who owns the underlying content you're reading.
Platforms don't claim ownership of your speech when you upload a video, either — YouTube's, TikTok's, and Instagram's terms all grant the platform a license to host and display what you upload, not ownership of it. So the rights question runs back to the creator, not the platform, in essentially every case.
The low-risk lane: personal use, notes, accessibility, research, journalism
A large set of common reasons to download a caption file sit in genuinely low-risk territory, and it's worth naming them explicitly because they cover most of what people actually do:
- Personal use. Pulling a transcript so you can search a 90-minute podcast for the one part you need, or paste a lecture into your own notes app, is private, non-distributive use. Nobody else ever sees it.
- Accessibility. Downloading captions to read along, translate for comprehension, or feed into a screen reader or accessibility tool is squarely within the purpose captions exist for in the first place.
- Research. Academic and independent researchers routinely extract transcripts to analyze language patterns, sentiment, or content trends — this is widely accepted, well-established practice, particularly when the output is analysis rather than the transcript itself.
- Journalism. Reporters quote from video sources constantly. Pulling a transcript to accurately quote a public statement, verify what was said, or fact-check a claim is a long-standing, broadly protected journalistic practice — most jurisdictions have specific carve-outs or strong precedent supporting news reporting and commentary.
- Note-taking for your own workflow. Summarizing a video for your team, your own reference, or a non-public working document is a different act than publishing the transcript itself.
None of this requires permission from the creator in the ordinary case, and none of it is meaningfully different from what people have always done with printed transcripts of speeches, interviews, and broadcasts — copy it down, use it privately, quote from it accurately.
The higher-risk lane: commercial and public republishing
Where things change is verbatim, public reuse — posting the full transcript of someone else's video on your own site, in your own newsletter, or as standalone content, especially where money is involved (ads on the page, a paid product, a business use).
This is higher-risk for a simple reason: you're not just reading the copyrighted work, you're distributing a full copy of it to a new audience, competing with (or substituting for) the original. That's exactly the kind of use copyright law exists to control. It doesn't mean it's automatically illegal — that's what fair use and fair dealing analyses exist to sort out — but it means you're now in territory where "I found it, so I can post it" is not a legal argument, and permission or a genuine fair-use argument matters.
The fair use / fair dealing test, in practice
In the US, fair use is a four-factor balancing test courts apply case by case — there's no bright line, which is part of why this area feels murky:
| Factor | What it asks | How it tends to cut | |---|---|---| | Purpose and character of use | Is the new use transformative (commentary, criticism, parody, analysis) or just a copy? | Commentary/analysis favors fair use; straight republishing doesn't | | Nature of the original work | Is it factual or creative? | Factual content (a lecture, a news clip) gets less protection than fiction or artistic work | | Amount used | How much of the work did you take, and how central is it? | Quoting a relevant excerpt favors fair use; posting the entire transcript weighs against it | | Effect on the market | Does your use substitute for the original, or hurt the creator's ability to monetize it? | If people can read your version instead of watching/paying for the original, that weighs heavily against fair use |
Outside the US, most countries with a common-law tradition (UK, Canada, Australia, and others) use "fair dealing" instead, which works differently — it's a closed list of specific permitted purposes (criticism, review, news reporting, research, private study — the exact list varies by country) rather than an open four-factor balancing test. If your use doesn't fit one of the named categories, fair dealing generally doesn't apply at all, which can make it stricter than US fair use in some ways and looser in others. The EU has its own separate framework of exceptions and limitations that again differs by member state in the details.
The practical read: transformative use — quoting a portion to analyze, critique, respond to, or report on — is on much firmer ground than posting a full transcript with nothing added. And because this genuinely varies by country and by the specific facts, "is this fair use" is exactly the kind of question worth a real answer from a lawyer rather than a guess from a blog post, if there's real exposure on the line.
Copyright law and platform Terms of Service are two separate questions
This is the part people conflate most often, and it's worth being explicit: whether reusing a transcript is legal under copyright law, and whether the way you obtained it followed the platform's rules, are two completely separate questions. You can be in the clear on one and not the other.
YouTube's, TikTok's, and Instagram's Terms of Service generally restrict scraping, bulk downloading, and automated access outside the way each platform intends its content to be accessed — regardless of what copyright law says about the content itself. Violating a platform's ToS is a contract issue between you and the platform (it can get an account suspended, or in some cases trigger a legal claim under something like the US Computer Fraud and Abuse Act for unauthorized access), not automatically a copyright violation. Conversely, using content in a way that's perfectly consistent with a platform's ToS can still infringe copyright if the underlying reuse crosses the line described above.
So think about these as two separate checklists, not one:
- Copyright: Am I reusing someone else's creative content in a way that needs permission or a fair-use argument?
- Terms of Service: Did I access or download the content in a way the platform's rules allow?
Platform-by-platform reality check
| Platform | Copyright sits with | ToS generally restricts | |---|---|---| | YouTube | The video creator/uploader | Automated scraping or bulk downloading outside YouTube's own API and intended use | | TikTok | The video creator/uploader | Scraping, automated data collection outside TikTok's intended use | | Instagram | The post/Reel creator | Automated data collection under Meta's platform policies, outside intended use | | X (Twitter) | The video poster | Automated access outside X's API terms |
The pattern repeats across every platform: copyright follows the person who made the content, and the platform's own rulebook is a separate layer restricting how you're allowed to pull data off their service, independent of what you're legally allowed to do with that data once you have it.
Attribution: not always required, always a good idea
Copyright law and attribution are not the same requirement — you can have full legal permission to use something (through a license, fair use, or the creator's own consent) and still have no independent legal obligation to credit the source. But attribution is worth doing regardless, for reasons that go beyond the legal minimum: it's the difference between "this looks like plagiarism" and "this looks like fair, honest reuse" to anyone who checks. It builds goodwill with creators instead of adversarial relationships. And in borderline fair-use situations, showing clear attribution and a link back to the original is one of the factors that can support a transformative-use argument — it signals you're pointing people to the source, not replacing it.
The practical habit: name the creator, link the original video, and note where the transcript came from, every time you reuse one publicly. It costs a sentence and removes an entire category of friction.
If a rightsholder objects: how DMCA takedowns work
If a creator or rightsholder believes their content — including a transcript derived from it — has been reused without permission, the standard mechanism in the US (and functionally mirrored by most major platforms worldwide) is a DMCA takedown notice under the Digital Millennium Copyright Act. The rightsholder submits a notice to the platform or host identifying the infringing content; the host is generally obligated to remove or disable access to it to keep its own legal safe harbor; the person who posted the content can file a counter-notice if they believe the takedown was mistaken or the use was legitimate (fair use, for example); and if the rightsholder doesn't pursue it further after a valid counter-notice, the content can generally be restored.
The practical implication: even a genuinely defensible fair-use republish can still get taken down first and argued about after, because the DMCA process is built around quick removal, not a pre-clearance legal ruling. That asymmetry is worth knowing before publishing a transcript in a way you'd have to defend.
A practical way to think about it
None of the above adds up to a single yes/no answer, because there isn't one — that's the nature of this area of law. But a reasonable, non-legal-advice mental checklist:
- Using it privately, for notes, accessibility, or research? Low risk, proceed.
- Quoting a relevant portion with commentary, analysis, or journalism attached? Reasonably solid ground, especially with attribution.
- Publishing the full transcript verbatim, publicly, especially where it's monetized? Higher risk — get permission, make sure your use is genuinely transformative, or talk to a lawyer before you do it at scale.
- Downloading in bulk via automated tools outside a platform's intended access method? Check the platform's ToS separately from the copyright question — it's a different rulebook.
One more time, because it matters: this page is general information, not legal advice. It doesn't account for your jurisdiction, your specific use case, or the actual facts of what you're doing. If there's real money, a business relationship, or genuine legal exposure involved, the right move is a conversation with a lawyer or, for accessibility-specific questions, an accessibility consultant — not a guide like this one.
Frequently asked questions
Is it illegal to download YouTube captions for personal use? No — downloading a transcript to read, search, take notes from, or use for your own accessibility needs is well within normal, low-risk use. The legal questions get more serious when you publish or redistribute that transcript publicly, not when you simply have a copy of it.
Can I get in trouble for using SubExtract or a similar tool to extract a transcript? The extraction tool itself doesn't create legal exposure — it's a mechanism, like a screenshot tool or a browser. What matters is what you do with the output afterward: private use is low-risk, public verbatim republishing is where the fair use and permission questions actually apply.
Does crediting the creator make reuse automatically legal? No. Attribution is good practice and can support a fair-use argument, but it isn't a legal shield on its own. You can credit someone accurately and still be infringing if the underlying use — copying the full work, competing with the original, no transformative purpose — doesn't hold up.
Is downloading captions against YouTube's Terms of Service? Manually viewing and copying captions through YouTube's own interface is normal use. Automated, bulk, or scraped access outside how YouTube's API and interface are meant to be used is what the Terms of Service generally restrict — that's a separate question from whether reusing the text is a copyright problem.
Do different countries really have different rules for this? Yes, meaningfully. The US uses an open four-factor fair use test; the UK, Canada, and Australia use fair dealing, a narrower list of specific permitted purposes; the EU has its own set of exceptions that vary by member state. The same reuse can be treated differently depending on where you and the rightsholder are, which is exactly why a specific situation deserves a specific legal opinion rather than a general guide.
Next steps
If you're actually pulling captions and want the mechanics rather than the legal landscape, download YouTube subtitles walks through the extraction step by step, and SubExtract's video captions tool is where that extraction happens. For what to do with a transcript once you have it — the low-risk, high-value side of reuse — repurposing YouTube content covers turning it into blog posts, newsletters, and social content the right way. And if journalism or research is your actual use case, SubExtract for journalists covers that workflow specifically.