IMPORTANT NOTICE FOLLOWING THE UNITED STATES COURT OF APPEALS, THIRD CIRCUIT OPINION (September 29, 2026, No. 25-2153)
- For reference https://www2.ca3.uscourts.gov/opinarch/252153p.pdf
- Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc, 25-2153, (3rd Cir. Sep 30, 2026) ECF No. 214
I will preface this that I am not a lawyer and thus only have my own personal non-legal opinion, however courts will take personal opinion and actions into consideration.
Considering in this opinion that usage of AI that has been deemed to not be in the fair use and that its use is not transformative, I am considering the following quote and my interpretation of it:
Second, ROSS argues that three "intermediate copying"
cases, one from the Supreme Court and two from the Ninth
Circuit, dictate a different conclusion. Opening Br. 44. In a
recent case, the Supreme Court held that Google could copy a
portion of Oracle's Java computer programming language to
enable users to create applications using Java code on Android
phones. Google LLC v. Oracle Am., Inc., 593 U.S. 1, 6-7
(2021). The Ninth Circuit concluded similarly in earlier cases.
It held that a video game developer could copy Sega's com-
puter program code to create video games that could be played
on Sega's gaming system. Sega Enters. Ltd. v. Accolade, Inc.,
977 F.2d 1510, 1514 (9th Cir. 1992). Another panel similarly
held that Connectix could copy Sony's computer program code
to create a product that made Sony video games playable on
personal computers. Sony Comput. Ent., Inc. v. Connectix
Corp., 203 F.3d 596, 599, 606-08 (9th Cir. 2000).
.
In each of these cases, computer code was copied to enable
users to make their software operable with an existing com-
puter system. These uses were transformative because, in each
case, copying was necessary to access the unprotected func-
tional aspects of computer code. See Oracle, 593 U.S. at 30-
31 (stating that Google copied Oracle API "only insofar as
needed" and explaining that "shared interfaces are necessary
for different programs to speak to each other"); Sega, 977 F.2d
at 1514 (noting that "no other means of access to the unpro-
tected elements exists"); Connectix, 203 F.3d at 599 (holding
that copying was "necessary to permit Connectix to make its
non-infringing Virtual Game Station function with PlayStation
games").8 Here, ROSS does not need to copy Thomson Reu-
ters's headnotes to access the underlying unprotected infor-
mation. ROSS had access to the underlying judicial opinions
and could freely copy them to make the memos needed to train
its AI. It chose not to do so because copying the headnotes
offered an "easy" way to create its training memos. App.
8545-46. Unlike necessity, ease is not a justification for cop-
ying. In sum, the first factor weighs against fair use.9
.
In conclusion, the second factor weighs slightly in favor of
fair use, but the first, third, and fourth factors weigh against it.
Evaluating these factors together, and in light of copyright's
purposes, we hold that ROSS's use was not fair. See UpCodes,
- Google v. Oracle directly affects SquirrelJME as it is a Java ME 8 Virtual Machine and Runtime Library which is a clean-room reverse engineered project solely based on documentation and observed behaviors of existing software.
- Considering that PhoneME, KVM, OpenJDK, and Android exist it would be impossible to determine if an AI has been trained on these projects or not.
- SquirrelJME is not a non-profit and has a Patreon along with sponsorships. Although I am not doing this with the intention of making money there is always the possibility of a non-MPL licensed SquirrelJME being purchased by an enterprise corporation.
- Google v. Oracle, Sega v. Accolade, and Sony v. Connectix are relevant to my own personal opinion as these specifically are in relation to interoperability with existing systems (i.e. emulation).
- With implementing SquirrelJME over the past decade, and considering the
existence of other projects such as the GNU Classpath for Java and
Apache Harmony. It would be impossible to argue that copying source code
from other projects in this manner is fair use. I have reasonably
demonstrated by documentation alone that you do not need to look at nor
reference any source code. This is ignoring
Randomand others such asString.hashCode()which must strictly follow the algorithm, thus its very few lined source code is printed in the documentation. That there only exists as Sun Microsystems wanted interoperability. - The court opinion does consider Generative AI vs Non-Generative AI; however what makes an AI generative is not something that has been determined in courts yet (it is a marketing term). Additionally, source-code specific AIs such as Copilot and Claude could very well end up being considered Non-Generative. Ignorance is not defendable in court, and it is a very well known fact that current AIs can regurgitate copyrighted code.
Therefor because of this, any use of AI is an existential threat to SquirrelJME and the project as a whole, thusly any and all usage is permanently banned until the appropriate time when this may be re-evaluated.
Future Policy Reservation
Due to court rulings and/or opinions, considering the time, cost, and lack of any existence of such AI technologies, this reservation is not being considered at this time.
Permitted use of GenAI with SquirrelJME with only authorized tools may be accepted in the future, if all the following conditions are met along with the main points of the contributing.mkd agreement:
- The GenAI models are solely trained on SquirrelJME's source code, notes, history, and documentation. They must not be trained on any other data which is not in the public domain.
- No generic models or additional models are applied, this includes additional models which provide support for human languages.
- It must not train from any other source that is not SquirrelJME.
- The use of the GenAI is run on the local device whether that be a CPU, GPU, or NPU without the use of an internet and network connection, it must be fully usable offline.
- The models must be fully reproducible and deterministic, the same input must result in the same output.
- The models must be able to be trained fully offline on the local device.
- The model training, prompt, and all other related software must be under an Open Source Initiative (OSI) license.
- All related software is able to be copyright, license, and trademark assigned to SquirrelJME in terms of full ownership with no restrictions. That is, it is donated to the project with no restrictions placed.
- The tool has been vetted and approved by the administrators of SquirrelJME.
Personal Thoughts
Personally disregarding legality of copyright and otherwise, working on SquirrelJME brings me much joy. One thing to consider is that GenAI is counter to the project's goal and desire to be autonomous and to act with self-determination. Programming is art, and SquirrelJME is made by humans for humans. Our passion and enjoyment is important.
Have you enjoyed making something? SquirrelJME is that enjoyment. Every line of code, graphic, test, input, and output that it happens brings me excitement and joy. One could just use OpenJDK or KVM to get effectively what SquirrelJME is, but that is not the point. Just getting code to run on your own virtual machine is extremely fascinating with lots of science stuff involved. This is something that I can nerd out about for many hours. Much of the world today is effectively just producing products out of an assembly line as quickly as possible and without any regards to quality. Think of yourself and what you enjoy, if that turned into an assembly line that pumps out content non-stop; think if you accepted forgetting what you love to do only to stand at said assembly line and tighten the same bolt when it comes up every five seconds. Would you enjoy that? I think that you would not.
Considering that courts are heavily involved now and the legal implications of these decisions and opinions, it would be foolish to destroy what SquirrelJME has for the convenience of not doing things yourself. It is now existential for the project, that is this technology cannot be adopted for the project to continue to exist. This by itself does not even consider the ethical issues of AI proliferation, which has repeatedly been shown to be damaging to the environment and the sheer disregard for the protection of creativity. In the current field of reverse engineering and emulation, AI has most popularly only been used for piracy and other black-hat projects. These projects have been a net-loss to preservation as a whole as they give courts and corporations more reason to squash these efforts by diluting projects that would legally stand on their own with copyright-stripped code from dubious sources. Current AI companies and users of AI are very aware of this and have adopted a no-care, ship it broken, and fix it in production mentality that only carries extreme risk. That is not something that I can do for this project which will soon reach 11 years of being in active development.
Risk has never been, nor ever will be acceptable for SquirrelJME.