Datasets:
Your unfounded DMCA takedown notices are an act of perjury.
If you submit a DMCA request to this repository after 2025-04-09T07:42:59, you have, under a reasonable reading of the law, committed an act of perjury.
Per 17 usc § 512 (which is the DMCA), any takedown notice requires:
- (iii) identification of the material that is claimed to be infringing or to be the subject of infringing activity and that is to be removed or access to which is to be disabled, and information reasonably sufficient to permit the service provider to locate the material.
- (v) a statement that the complaining party has a good faith belief that use of the material in the manner complained of is not authorized by the copyright owner, its agent, or the law.
- (vi) a statement that the information in the notification is accurate, and under penalty of perjury, that the complaining party is authorized to act on behalf of the owner of an exclusive right that is allegedly infringed.
The dataset was removed entirely from HuggingFace at 2025-04-09T07:42:59. After this point, you could not have viewed the data. Thus:
- You could not, in good faith, identify specific infringing material (c3A iii),
- nor could you affirm a good faith belief in infringement (c3A v),
- and any declaration of accuracy (c3A vi) would be knowingly false.
Submitting a DMCA takedown notice without verifying the material constitutes perjury and exposes you to liability under 17 usc § 512 (f).
Further:
- You can not reasonably request for the dataset to be taken down. To do so is equivalent to asserting that you hold copyright for every work in the dataset, which you do not.
- The OTW does not hold copyright in individual AO3 submissions and cannot issue mass takedowns.
Courts have consistently punished abusive DMCA practices:
- Lenz v. Universal Music, 2016— If you fail to consider whether the material you are submitting a takedown notice for is covered under fair use, you cannot possess a good faith belief of its infringement, and thus are violating 17 usc § 512 (f).
- Automattic v. Steiner, 2015 — False takedown notices cost defendant $25,084 in damages and fees.
- Online Policy Group v. Diebold, 2004 — Misuse of DMCA notices led to monetary liability.
- Oppenheimer v. Allvoices, 2014 — Negligent identification of copyrighted work is actionable by a court.
The OTW's original takedown notice was unfounded and likely illegal. Per their website:
The Organization for Transformative Works does not hold the copyright to your fanworks, so we cannot contact other sites or organizations on your behalf. Because you are the copyright owner, you will have to contact the other site yourself to request that they remove the unauthorized copy of your work.
Yet, they issued such a request anyways that was very definitely legally actionable. Even if the OTW decides not to pursue legal action, they are still vulnerable to a huge amount of liabilities right now.
TL;DR:
- Submitting a takedown request is not protest or activism. It is a legally actionable move, and you can be sued for it.
- You do not need to care whether it's fair use or copyright infringement to still be in the legal wrong for submitting a DMCA notice.
- The OTW broke the law in submitting their takedown request, and would look insane to a judge if they tried to claim copyright over your works.
Stop doing this.
Hey everyone, just wanted to flag this post as potentially misleading and a scare tactic. As individual creators, we absolutely have the right to file DMCA takedowns for our own copyrighted works that we find posted without our permission. The user in that post is attempting to apply legal arguments about potential misuse of the DMCA system to individual creators acting in good faith to protect their own stories. The scenarios are different, and their conclusions are likely incorrect in your case.
Official information or policy changes regarding DMCA processes would almost certainly come directly from Hugging Face staff, not from a random user in the discussions.
Key points to remember:
- You know your own work. If you find it posted somewhere without your consent, you can identify it.
You have a good faith belief in infringement when your copyrighted work is used without your permission. - As many of you have noted, the presence of AO3 IDs on the works within the dataset description provides clear evidence that our content was taken and reposted. This directly contradicts the claim that you couldn't identify your work.
- Filing a DMCA for your own work to protect your rights is generally a legitimate use of the DMCA process.
Don't let this post discourage you from taking necessary steps to protect your creations. If you have concerns about a DMCA you've filed, it's always best to seek clarification from official sources or legal counsel if needed.
Let's be careful about spreading unverified information. If Hugging Face has any official announcements regarding DMCA policies, they will communicate them through their official channels. Please stay vigilant!
Does anyone know where you can report fraudulent DMCA takedown notices, or if you have to be the one named in the DMCA takedown notice to report a DMCA takedown notice as fraudulent?
Also, do their posts on the discussion tab even constitute a DMCA takedown notice? I was under the impression that HuggingFace had separate methods of contact for DMCA takedowns.
Thanks, I'll even pin it
As individual creators, we absolutely have the right to file DMCA takedowns for our own copyrighted works that we find posted without our permission.
Sure, but fair use is a notable exception. There is absolutely a significant legal standing to argue that this dataset constitutes fair use.
The user in that post is attempting to apply legal arguments about potential misuse of the DMCA system to individual creators acting in good faith to protect their own stories. The scenarios are different, and their conclusions are likely incorrect in your case.
The court cases I listed are not direct parallels. That was never claimed. See the text directly above that list where I explained the purpose of its inclusion.
Official information or policy changes regarding DMCA processes would almost certainly come directly from Hugging Face staff, not from a random user in the discussions.
HuggingFace is not an authority on DMCA and this is a matter of law, not site policy.
You know your own work. If you find it posted somewhere without your consent, you can identify it.
It would not have been possible for someone to find their works in this dataset past 2025-04-09T07:42:59 and thus could not identify it.
You have a good faith belief in infringement when your copyrighted work is used without your permission.
No you do not, not when you have not additionally verified the use of the work does not constitute fair use. This is the exact reason your fanfiction is not illegal.
As many of you have noted, the presence of AO3 IDs on the works within the dataset description provides clear evidence that our content was taken and reposted. This directly contradicts the claim that you couldn't identify your work.
The dataset is not and does not claim to be exhaustive. The description presents detail about how the data was gathered, not a guarantee of its contents. That your work's ID was in a certain range does not mean it was included in the dataset, and there are plenty of works in my scrape of AO3 that are not in this data.
Don't let this post discourage you from taking necessary steps to protect your creations. If you have concerns about a DMCA you've filed, it's always best to seek clarification from official sources or legal counsel if needed.
I am legal counsel, and I am telling people to not be manipulated into submitting legal requests they don't fully understand that expose them to liability.
Let's be careful about spreading unverified information. If Hugging Face has any official announcements regarding DMCA policies, they will communicate them through their official channels. Please stay vigilant!
HuggingFace does not make the law and has no say in whether I sue you.
If you were the creator of a resource targeted by a fraudulent DMCA notice, you can submit a counter-notice pursuant to the relevant law linked above and then sue. Otherwise, you may just sue. Have fun ^_^
wouldnt it be a nightmare if we lived in the world where the metric of how angry a group of people got in the world actually dictated and changed the fundamental rules involved in how legal proceedings are carried out?
"it should be illegal" =/= "it is illegal"
if you feel strongly about this, you are allowed to feel that way (and that is your prerogative as an author), but as of April 2025, that is pretty clearly Not How The Law Works™
there is also a subconscious bias going on here where the takedowns are flagging a dataset that has been exclusively used in research or ML hobbyist settings rather than for corporate production models; i want to make it viscerally clear to the people reading this that companies like Anthropic/OpenAI do this aggressively, privately, and explicitly for the purposes of making money, while HF as an org predominantly is associated with people who are interested in independent research, academia, or hobbyist experimentation of these models, which is why this dataset is public to begin with. basically it's an "out of sight out of mind" kind of bias, where people get angry at the only org that has people who are legitimately concerned about sharing reproducible and legitimate research.
from my perspective, this seems sort of like protesting big pharma by bringing a flamethrower to a clinic in your town or something. this only hurts people who are doing what they are doing downstream of the existence of large AI companies who aren't being transparent about their data practices to begin with
wouldnt it be a nightmare if we lived in the world where the metric of how angry a group of people got in the world actually dictated and changed the fundamental rules involved in how legal proceedings are carried out?
"it should be illegal" =/= "it is illegal"
if you feel strongly about this, you are allowed to feel that way (and that is your prerogative as an author), but as of April 2025, that is pretty clearly Not How The Law Works™
there is also a subconscious bias going on here where the takedowns are flagging a dataset that has been exclusively used in research or ML hobbyist settings rather than for corporate production models; i want to make it viscerally clear to the people reading this that companies like Anthropic/OpenAI do this aggressively, privately, and explicitly for the purposes of making money, while HF as an org predominantly is associated with people who are interested in independent research, academia, or hobbyist experimentation of these models, which is why this dataset is public to begin with. basically it's an "out of sight out of mind" kind of bias, where people get angry at the only org that has people who are legitimately concerned about sharing reproducible and legitimate research.
from my perspective, this seems sort of like protesting big pharma by bringing a flamethrower to a clinic in your town or something. this only hurts people who are doing what they are doing downstream of the existence of large AI companies who aren't being transparent about their data practices to begin with
so like, legitimate question, but do you let your doctor perform wild medical research on you without your consent? 'Cause the biggest flaw in your logic is people participating in research studies are, generally, aware they're participating in a research study and consent to it
also, 2nd legitimate question, but by this logic, you're agreeing that omegaverse mpreg yaoi fanfiction is an integral part of academia?
