CIPPIC
has filed an interesting and apparently unprecedented Federal Court application claiming standing as
an “interested person” with respect to a Canadian copyright registration for a
work entitled SURYAST in which an AI program is named as a co-author. CIPPIC says on its website:
“CIPPIC is
asking the Federal Court to expunge or rectify a copyright
registration that holds out an artificial intelligence (“AI”) program as
an author under Canadian copyright law.
At the heart of
this dispute is SURYAST, an AI-generated output, that raises fundamental
questions about authorship, originality, and AI’s place in creative endeavors.
Ankit Sahni,
an intellectual property lawyer from New Delhi,
obtained a copyright registration for SURYAST, listing himself and the AI
system as co-authors. The SURYAST output is an adaptation of a sunset
photograph, taken by Mr. Sahni, but distorted by the AI according to the style
of Vincent van Gogh’s “The Starry Night.””
Here’s the work in question entitled “SURYAST”:
(SURYAST
(Hindi word for sunset); generated with the assistance of RAGHAV) From the CIPPIC website.)
Here’s the CIPO (“Canadian
Intellectual Property Office”) copyright registration, which lists the “authors”
as “RAGHAV Artificial Intelligence Painting App” and Ankit Sahni. Mr.
Sahni is listed as the owner. The “authors” and owner are all at the same
address in India.
CIPPIC says in its application memorandum
that:
[3] The
Respondent created a “legal test case” by exploiting the automated registration
system of the Canadian Intellectual Property Office (CIPO) to obtain a
copyright registration for the SURYAST output. CIPO duly issued the
registration without examination, identifying the Respondent and his AI
application as co-authors. That registration is now cited internationally as evidence
that Canada recognizes AI authorship, creating a misleading precedent that
harms the integrity of the public register and muddies the copyright status of
AI outputs in Canada.
For his part, the Respondent Ankit Sahni says in his
application memorandum:
7. The
Respondent accepts the description of the background facts as set out by the Applicant.
CIPPIC says in a letter in its record from
David Fewer, its Director and General
Counsel, to the CEO of CIPO, citing an article from an Indian law firm,
about the Canadian registration:
The
registration as it stands is improper and raises significant concerns about the
overall accuracy of the Register of Copyrights. It also sets a problematic
precedent for Canadian copyright law. It opens the door for other registrations
with AI “authors” that is contrary to the guiding principles of Canadian
copyright law and undermines the entire legislative scheme. Canada is being
pointed to internationally as a jurisdiction that “recognized Sahni’s
co-authorship with the AI tool”.
This is, with respect, neither
factually nor legally accurate. No Canadian legal precedent has been set. Copyright
registration in Canada involves no examination and no “deposit” of the work in
question. If I want to register my supposed copyright interest in a roll of
toilet paper with Donald Trump’s picture and name on it that I created with AI and
call it “Tribute to Trump”, I can do so online and waste $63. I can list myself
as author. The CIPO computer will happily process my payment and not ask any questions.
CIPO will not look at the “work” and will not even allow me to submit it. “Certificates
of registration may not be strong evidence of the subsistence of copyright or
its ownership, but that is a question of weight.” See Patterned Concrete Mississauga Inc. v. Bomanite Toronto Ltd., 2021 FC
314 (CanLII), <https://canlii.ca/t/jf8r0>
para. 13. So, it’s not hard to flush that evidence away with stronger evidence.
This shows what information
the Canadian Government requires for registration of copyright in Canada. It
says:
Do not send a copy of your work along with the
application. The Copyright Office does not review or assess works in any way,
nor does the Office check to see whether the title of your work has already
been used. Many works may appear with the same title, but if each work has been
created independently, each will have its own copyright protection.
(highlight added)
As CIPPIC points out, copyright
registration in Canada is provided “automatically”. According to CIPPIC:
“[13] The
Respondent characterizes his copyright registration efforts for the SURYAST
output as “legal test cases.” On his public LinkedIn profile, the Respondent
states he has been “working on several legal test cases, including being the
first person to have an AI system recognized as a co-author of a copyrighted
work in India and Canada.” [FN omitted]
[14] The
Respondent sought copyright registration for the SURYAST output in Canada, the United
States, and India. India’s Copyright Office initially registered the work and
recognized RAGHAV AI as a co-author, but later issued a withdrawal notice.
However, the registration remains publicly listed. The United States Copyright
Review Board rejected the application, holding that the SURYAST output lacked
sufficient originality and that an AI system cannot be an author. [FN
omitted]
The American copyright registration
system is totally different than the Canadian system. It does involve
examination and deposit. To compare it to that of Canada is like comparing Notre
Dame cathedral to the tool shed in my backyard. The Canadian system is,
frankly, an embarrassment and serves no function sufficient to justify its
continued existence. It
could have been used, by requiring registration and a
non-trivial fee for the extra 20 years, to mitigate the unfortunate Canadian decision
to extend the copyright term to life + 70 years
– but the Trudeau government unfortunately failed to do this.
The US Copyright Office also plays an
important and normally well-informed role in domestic and international policy
– even if this sometimes sparks controversy. BTW, here’s the latest report from Shira
Perlmuter, the US Register of Copyrights and Director of the U.S. Copyright
Office – who is so far successfully resisting the efforts of Trump to fire her.
The US Copyright Office has taken the position
that AI cannot be an “author”. “Because an AI system is not a human being, it
cannot be considered an “author” in collaboration with a user.” See also this commentary.
CIPPIC has filed a very lengthy
349 page record in this case complete with “expert” evidence. The
Respondent, Ankit Sahni, “accepts the description of the background facts as
set out by the Applicant.” Here is Mr. Sahni’s record.
The sometimes influential Canadian Music
Publishers Association (“CMPA”) has filed a 74 page motion to intervene and
apparently supports CIPPIC’s position that an author must be human.
Neither Mr. Sahni nor CMPA seem to have
any problem with the nature of this proceeding or the standing of CIPPIC as an
“interested person” to initiate it. Indeed, CMPA supports CIPPIC’s contention
that it is an “interested person”, which should be the existential threshold issue
in this proceeding, as I discuss below.
Frankly, one cannot help but wonder how
this case arose or to what extent it is actually adversarial in nature. If Mr.
Sahni loses, he will have something more to write about in his apparent quest
to publicize his test of various registration regimes. In the unlikely event
that anyone could ever be bothered to infringe SURYAST and Mr. Sahni decides to
sue in Canada, he could still
conceivably try to do so successfully as an author/owner who used AI as a tool.
If CIPPIC somehow “wins” in this application, in spite of the “standing” issue
discussed below, this registration might
possibly be struck – but the ruling could and perhaps should be quite narrow
and procedural and provide no useful precedential guidance.
If and when Mr. Sahni sues someone in a
Canadian court for infringing SURYAST, then CIPPIC could then apply to
intervene, as it has done in many other cases.
“Interested Person”
CIPPIC may risk the displeasure of the Federal
Court by claiming to be an “interested person” in this case. The only possible
basis for CIPPIC’s application is that it is “interested person” within
the meaning of s. 57(4) of the Copyright Act. Frankly, its arguments are
unconvincing. If CIPPIC is an “interested person” in this instance, so too are
countless Canadians and others who may, though lacking CIPPIC’s experience, have
well informed views about AI and copyright law. Being an “interested person” is
not the same thing as being an intervener, a role in which CIPPIC is
indeed experienced – and which requires leave of the Court in any given case.
Being an “interested person” is not the same thing as finding an issue to be
interesting, however interesting and even important it might be.
S. 57(4) of the Copyright Act
states that:
57(4) The
Federal Court may, on application of the Registrar of Copyrights or of any interested person,
order the rectification of the Register of Copyrights by
(a) the
making of any entry wrongly omitted to be made in the Register,
(b) the
expunging of any entry wrongly made in or remaining on the Register, or
(c) the
correction of any error or defect in the Register,
and any rectification of the Register
under this subsection shall be retroactive from such date as the Court may
order.
(highlight and
emphasis added)
It would seem quite questionable that
CIPPIC is an “interested person”. The case law under s. 57(4) suggests that
an “interested person" must be an author, owner, someone disputing or
updating ownership, plaintiff, defendant, assignee, licensee, or someone
claiming to be any of the foregoing, etc. This case law is not to be found in
CIPPIC’s Memorandum. Mere curiosity or intellectual interest is arguably far
from sufficient. That kind of interest is best pursued in academic journals,
blogs, or lobbying corridors. There is an old proverb about how “curiosity killed the cat”.
It would seem that CIPPIC, Mr. Sahni,
and now the CMPA are all eager to bring a potentially very complicated test
case to the Federal Court – even though there is no “live dispute” between the parties involving a
potential finding of liability, or any basic factual dispute. It is interesting
that the records do not indicate that there was any cross-examination on the
various affidavits.
Let me be blunt. IMHO, This is not an
appropriate case to take to the Federal Court. The Court may not be willing to
spend a lot of time and resources in these circumstances, even if the parties
and intervener(s) are willing to expend their resources in this way. CIPPIC is not defending anyone who has been sued.
The Supreme Court of Canada has made it crystal clear that courts should not be
granting declaratory judgments or advisory opinions in copyright cases where
there is no actual “live dispute”:
But the usefulness of guidelines in
theory does not provide
the Court with a sound basis for entertaining declaratory relief without a live
dispute between the parties or when those whose rights are at stake
are not privy to the proceedings.
(highlight and
emphasis added)
See York University v. Canadian Copyright
Licensing Agency (Access Copyright), 2021 SCC
32 (CanLII),
[2021] 2 SCR 734, para. 86. Certainly,
the long running confrontation between Access Copyright and York University
that led to this landmark decision began as a very real “dispute” with
countless millions of dollars actually at stake for York and other educational
institutions in stark comparison to the theoretical debate between CIPPIC and
Mr. Sahni. But since the tariff in question was ultimately held to be “not
mandatory”, there was no longer any actual “live dispute" between the
parties. Accordingly, and thankfully, the Court declined to rule on the very
problematic guidelines upon which York tried to rely. BTW, I’m very proud that
my arguments on behalf of CARL and Prof. Ariel Katz’s arguments were
instrumental in carrying the day on the “mandatory” issue when the case got to
the SCC.
The issue of whether a public interest
entity has standing to ask the courts to rule in the absence of an actual live
dispute or as other than a directly affected person in an actual case has been
and continues to be important, and probably always will be. CIPPIC’s relies on Canada
(Attorney General) v. Downtown Eastside Sex Workers United Against Violence
Society, 2012 SCC 45 (CanLII), [2012] 2 SCR 524, <https://canlii.ca/t/fss7s>.
However, CIPPIC can hardly compare its interest here to that of the organization
representing the sex workers whose safety and business interests were obviously
at stake in that litigation but who were unwilling themselves to bring a challenge
forward in their own names. See also Borowski
v. Canada (Attorney General), 1989 CanLII 123 (SCC),
[1989] 1 SCR 342.
CIPPIC boldly submits it has that it
has standing as an “interested person”:
[39] No private
party is directly interested in challenging the SURYAST output registration.
The work does not infringe the rights of any identifiable person or entity.
There is no private squabble over copyright ownership. This is not a defect in
CIPPIC’s application, but a reason why CIPPIC merits public interest standing.
The Respondent’s registration raises a systemic question about the validity of
non-human authorship under Canadian law and the subsistence of copyright in AI
generated outputs. These questions will go unaddressed unless this Court allows
a party like CIPPIC to bring them forward.
CIPPIC’s reliance on Rizzo &
Rizzo Shoes Ltd (Re), [1998] 1 SCR 27, <https://canlii.ca/t/1fqwt>
(SCC) to expand the meaning of “interested person” is anything but persuasive.
That case is too frequently cited as a last resort when no better argument or
authority regarding statutory interpretation is available.
There’s lots of other case law about
public interest standing – and it’s a very live issue even as we speak in the
SCC, i.e. the recently heard Democracy Watch v. AGC case in which
judgment is pending. It’s obvious that
the courts can and must exercise some serious triage on when and how
non-parties can initiate litigation. Canadian courts are notoriously backlogged
with urgent real litigation while resources, especially judicial resources, are
and are likely to remain limited and even inadequate overall.
BTW, I’m not the only one to raise the
issue of whether CIPPIC has “standing” here. See this comment from the prominent firm of Lenczner Slaght.
Other Issues
CIPPIC is seeking:
an Order:
i. pursuant to
paragraph 57(4)(b) of the Copyright Act, to rectify the Register of Copyrights by expunging the
Registration dated December 1, 2021, in connection with the SURYAST output
(Canadian Copyright Registration Number 1188619); or
ii. in the alternative, pursuant to
paragraph 57(4)(c) of the Copyright Act, to rectify the Register of Copyrights by removing “RAGHAV
Artificial Intelligence Painting App” from the SURYAST Registration as a
co-author;
(highlight
added)
For his part, Mr. Sahni, through his
counsel from a very reputable IP firm, makes some frankly bizarre assertions
such as:
25. While the
Court may opt to exercise its discretion, the Registrar is entitled to an appropriate degree of
deference. (FN omitted)
26. The Registrar has determined that rigorous examination of
applications for registration is unnecessary. Unlike other intellectual
property such as patents and trademarks, copyright law is sui generis.
(highlight
added)
There no longer appears to even be an
actual “Registrar of Copyrights”, although I recall that there once was such a
person. The term still does appear in the Copyright Act. In any case, the
“Registrar” (or whoever or whatever) is not entitled to any deference because nothing
was done or decided whatsoever other than that the CIPO computer automatically accepted
Mr. Sahni’s $63 and automatically issue a certificate in accordance with an essentially useless registration regime
that is predicated upon NO examination and NO deposit.
Mr. Sahni’s counsel states:
28. The simple purpose of copyright registration is to provide
official proof of ownership for creative works. Indeed, in Canada, copyright registration
automatically exists upon creation. Registration simply provides a legal
certificate to be held by the owner to protect against infringer.
(highlight
added)
The $63 certificate provides nothing
more than minimal and evidence pursuant to s. 53 of the Copyright Act
that can easily be challenged and overcome in litigation, depending on the
facts. That is hardly “official proof of
ownership”. It is simply incorrect to state that “copyright registration
automatically exists upon creation.” Obviously, registration does not happen
“automatically”. Someone must create an
account with CIPO, provide minimal information, and – above all – pay $63 (or
$81 if done in writing).
Sahni’s memorandum does contain some
interesting observations about modern art created by throwing paint on a
canvas, etc. This is NOT the kind of thing that should take the Court’s time in
this way.
AI does not create works all by itself,
although that may soon happen or even be happening. That is not alleged in this
case. Mr. Sahni had to prompt the machine, even if only with some minimal
parameters. The Supreme Court of Canada has made it quite clear that “skill and
judgment” are required for the subsistence of copyright in Canada. It may well
be that AI can be used by human authors to assist in the production of works
protected by copyright. Various “machines” and technology have been used since the
advent of Gutenberg’s printing press in 1440 and particularly in the last two
centuries to enable the production and reproduction of copyrighted works
ranging from books to photographs to films to musical compositions to
performances. AI is obviously a new and very promising and potent tool.
If and when Mr. Sahni should ever sue
someone in Canada for infringing SURYAST, maybe we will find out whether the
work actually qualifies for copyright and whether even he qualifies as the
author. He won’t need to join the AI
program as a co-author - unless he wants to generate more publicity and potentially
unnecessarily spend a lot of his and others’, including the Court’s, time.
Is the Issue Really So Clear?
The ultimate issue that CIPPIC has
exposed is whether AI can be an “author” for copyright purposes. One is tempted
to categorically deny such a possibility and to look somehow for legal
confirmation. However, remember the old “infinite monkey theorem” which
“states that a monkey hitting keys independently and
at random on a typewriter keyboard for
an infinite amount of time will almost surely type any
given text, including the complete works of William Shakespeare”.
Maybe the issue isn’t as simple and obvious as it seems. Given
the appalling and often minimalist state of much of today’s visual and musical
production, for example, I have no doubt that AI programs, left to their own
machinations, can produce much better stuff than we often see in museums or
hear on Spotify. What if AI is able to generate, without much or even any
further human involvement, more Mona Lisa style paintings, more Shakespeare
style plays, and more Mozart style symphonies – all of which
might well be quite new and different than anything before, technically and
qualitatively excellent, and very “creative”?
Perhaps the issue involves not only whether AI can be a an
“author” in the co-author sense, as in this application, but an author all by
itself without any specific human prompting or further involvement. What do
“author” and “creative” really mean anymore?
Is it maybe even timely and necessary to revisit the meaning
of “original” in copyright law? Can something be “original” and maybe even
magnificently so without human authorship? Canada’s iconic CCH v. LSUC decision defined “original” as follows:
25 For
these reasons, I conclude that an “original” work under the Copyright Act is one
that originates from an author and is not copied from another
work. That alone, however, is not sufficient to find that something
is original. In addition, an original work must be the
product of an author’s exercise of skill and judgment. The exercise of
skill and judgment required to produce the work must not be so trivial that it
could be characterized as a purely mechanical exercise. While creative works
will by definition be “original” and covered by copyright, creativity is not
required to make a work “original”.
See “The Evolution of Originality in Canadian Copyright Law:
Authorship, Reward and the Public Interest” Carys J. Craig,
2005. Perhaps “the times, they are a-changin”…?
This is fascinating
and interesting stuff – but I’m skeptical that it should be justiciable
as framed in this application which is focussed on one isolated registration in
Canada’s problematic copyright registry involving a work that is not the
subject of any “live dispute”. If this case somehow goes forward, it may and
perhaps should be more about standing and procedure than about the potentially
very complex issues of authorship, originality, creativity, etc.
Conclusion
CIPPIC has done some very good work
over the years. Indeed, I was proud to have been CIPPIC’s pro bono external lead
counsel on its first and possibly most important court case where we successfully
and influentially intervened to protect people’s privacy rights in the face of
mass music industry piracy litigation. See: BMG Canada Inc. v. John Doe
(F.C.A.), 2005 FCA 193 (CanLII), [2005] 4 FCR 81, <https://canlii.ca/t/1kx1k>. CIPPIC has also intervened successfully in other
important cases, e.g. involving default judgments in more recent mass piracy litigation,
reverse class actions, and lately preserving fair dealing rights in the face of
a dubious TPM circumvention allegation (appeal pending).
This is, frankly IMHO, a needlessly
risky initiative for CIPPIC in the sometimes rather harsh real world of the Canadian
Federal Court. It is clear that much of the work was done by students. Student
involvement in actual litigation is commendable and should be encouraged in the
right circumstances. But perhaps this case should have been framed as a law
school moot court exercise based upon a hypothetical fact situation – rather
than an actual Federal Court application that could consume a lot of precious
court time potentially only to be dismissed for lack of standing and/or absence
of an actual “live dispute” involving actual parties. That would conceivably be
a big disappointment for the students involved.
For whatever reason, this case seems to
be going nowhere according to the docket. A Requisition for
Hearing was filed on July 28, 2025 but nothing appears to have happened since
then. CMPA’s application to intervene has not been decided. Might this be
because the Court perhaps has some concerns about this application? If it does
somehow proceed, and there is no further intervention to challenge the basis of
the proceeding, including CIPPIC’s assertion of standing, it will be
interesting to see if a sceptical judge might do so sua sponte at some
point.
At the end of the day, CIPPIC has admittedly identified an
interesting issue. But this proceeding is NOT the best, maybe not even a viable
way to resolve the issue, and possibly very risky for CIPPIC. It The issue of
whether AI can be an “author”, as opposed to being yet another tool used by
actual human authors, may get resolved one day in Canada through actual
adversarial copyright infringement litigation. If the issue is sufficiently
serious and urgent, which is currently far from clear, then the Government could
legislate. As for registration of a non-human as an “author”, the Government
could presumably deal with this issue by regulation. If and when the Government
chooses to engage in significant broad revision of the Copyright Act, or
even in updating regulations under that legislation, this is an issue that
might well be addressed at that time.
And finally, for those interested in
this issue in whatever way, stay tuned for what is happening in the USA where
the Thaler case may be heard in the US Supreme Court if certiorari is
granted. Here’s the US Government’s recent brief
on “Whether the court of appeals correctly upheld the Copyright Office’s
refusal to register a claim to copyright in an image for which no human author
had been identified.” Here’s the docket in
the Thaler case to date. Here’s the decision from the DC
Court of Appeals.
BTW, the University of Ottawa Faculty
of Law is once again looking for a new Director of CIPPIC.
HPK
(highlight and emphasis added)
(as usual, not legal advice)
Posts
No comments:
Post a Comment