Panel 2 – Copyright Enforcement:
Faye Fangfei Wang, Resolving Copyright-related Cases Over the Internet
with the Assistance of Artificial Intelligence in Europe
Automated notice and takedown/Content ID with appeal mechanism as an
example of how the new European rules are supposed to work. Automated
mechanisms are supposed to be used for identification but not legal assessment.
Conflict over whether this is a general monitoring obligation. Proposed Artificial
Intelligence Act requires assessment of AI to ensure that it doesn’t distort
behavior or cause physical or psychological harm: this interacts with the ©
regime. Identifying audio/video is a challenging task where there are changes
in speed, background noise, etc. Use as part of a lecture, video game being
played, etc. Risk assessment/risk scale should be part of the integration—degree
of confidence in detection should matter to treatment.
Cathay Smith: Do you foresee a point in time at which an AI might be able
to make a fair use/fair dealing decision in the moment? Consider example of
police officers playing music to deter online posting of video of their actions—exploiting
the automated systems.
A: not ready yet. But fair use/dealing typically involves reproduction of
a limited portion, and AI could calculate portion. This may be one factor.
Identifying reason for reproduction may be far harder for an AI—commentary or just
RT: “Limited portion” is more true with video and songs, less true with
photography. The European regime doesn’t seem to have given any thought to how
this will work with photography; does AI make distinctions? Also interested in
reactions to German implementation proposals which seem to have some thresholds
attached, not just sliding scale.
A: Content ID is video and audio; AI may be able to match pictures.
Elizabeth Townsend Gard, Creativity in the Shadow of the Case Act
creators focused on quilting and copyright; online copyright camp for creators,
mostly women—educating creators on nuances is empowering and they spread the
world. Disrupted by CASE Act. Reviewing the comments on CASE Act: the comments
were focused on procedure. EFF doesn’t want fair use cases in the system; MPAA
does. Comments demand ease and accessibility but the discovery rules are
insanely complex, which is a concern. There is fear and threats surrounding
opting out on both sides. Librarians are threatening people: you might have
bigger fees; some tell patrons not to opt out. Should there be public lists of institutions
that plan to opt out so that people don’t waste the fees on filing? There’s no
conversation about creativity. CASE Act is detracting from education about the
fundamental principles of copyright.
In comments: Songwriters
say that they can’t afford to file cases if there will be opt-outs, despite
their activism in pushing for CASE in the first place. Science Fiction authors:
won’t be useful for SWFA. The creators themselves are saying it’s not for them.
Big corporations say they’ll opt out; libraries the same. Then who is this for?
How do you
separate TM and ©? The disputes often overlap. The CASE Act doesn’t seem to
RT: Who is it
for? I think it’s for photographers. I wonder what it would be like if it had
been written just for photographic works.
A: Is hearing
other creators say “it’s my time for revenge!” and also people worried about
liability so they may not opt out. It’s the uncertainty that is so difficult.
Karshtedt: Overlap with constitutional issues in patent: resolution of claims
in administrative proceeding. The policy issues you raise are different but
perhaps there’s an interaction with the constitutional issues.
A: The comments
raise the constitutional issues. It wasn’t really fully baked when it passed
and the fact it went through Dec. 20 as part of the Covid package created
problems. Putting David Carson, with a track record with the Office, in charge suggests
they really want it to work, but it’s still half-baked in terms of
Karshtedt and Sean Pager, Volition and Intent in the Law of Direct Copyright
Patent and ©
treat these areas very differently. Advent of new tech changed focus from
direct to indirect, and back to direct again. Congress amended the statute to
ensure that retransmissions of cable would be public performance; then the
issues shifted to copying. Netcom raised the problem of volition for copying
for online servers. Aereo brought direct infringement back into focus where
Netcom had seemed to settle the issue; put pressure on the line between direct
and indirect liability. Netcom said that volition was lacking when the user was
the one who made the copy and the service was a passive conduit and could not
be directly liable though indirect liability was possible in theory; it was
almost a moral distinction. Aereo complicates that. Cable-likeness established
its direct liability. It’s unclear if there’s a direct infringer, which matters
because there can’t be indirect liability without someone else directly liable.
Loosely similar to divided infringement in patent law where no one person
infringes but their conduct together does.
Circuit thinks that volition is still an element; SDNY courts have held in the
embedding context that volition is not an element after Aereo and embedding can
be infringing display.
was focused on Aereo’s lack of curation, but they don’t think that matters.
Indirect liability depends on the associated circumstances. If you set up a webcam
in a coffee shop, and someone happens to sing a song that gets transmitted, you
are responsible for the resulting public performance even if you lack knowledge
that it happened. Thus, even if the majority’s cable-likeness test is
unsatisfying, Aereo’s setup was directly responsible for the public
performance, which justifies liability in this case. Cablevision has a much
more elegant analysis: asks whether contribution to creation of infringing copy
is so great that it warrants direct liability even though another party has
made the copy. For Aereo, only the tech entity is responsible for the public
performance. Both majority and dissent in Aereo agree that on-demand providers
perform. SDNY opinion on display gets to the same result through proximate
causation, which they don’t think is right—causal responsibility is different.
forward: a causal responsibility approach to direct liability. Rooted in moral
philosophy, criminal/tort law about innocent instrumentalities.
have a purely passive role; functions automatically at behest of user. Aereo
could be said to be supplying a product, but Aereo has the ability to control.
Simplistic to say that only button-pusher can be a direct infringer. Did the
defendant play a substantial enough role in the user’s acts to become causally
responsible? Did defendant control the circumstances in which the user
performed the acts in question? Aereo transmit content on demand, and the user
triggers that by pushing a button. In the aggregate, this is infringing public
performance because Aereo is causally responsible.
So we think
that public display SDNY results are correct because D are causally
responsible. But Usenet simply provides access, which is not enough for causal
responsibility, so 9th Cir. Giganews can be right too (Perfect 10?).
Matt Sag: intent?
exists as an element but not w/r/t specific works, just general intent that
there will be a performance. If your whole service is designed so that content
can be transmitted to people, then hard to deny knowledge that it will get transmitted.
[That sounds a lot like indirect liability to me. The underlying idea is that
all or most of what gets transmitted will infringe, and that’s a standard
RT: (1) Perfect
10? (2) Haven’t you just shoved the Aereo question into the phrase “in the
aggregate” because the right is public performance, not performance, so we have
to decide what counts as a performance to the public, so that you’re assuming
the conclusion? Compare Dropbox which happens to back up multiple copies of The
Big Lebowski, each downloaded by a different iTunes user that has legitimately
paid to do so?
A: it’s a
different kind of intent. In the public performance cases you do have to answer
the aggregate question, which goes back to the transmit clause which defines
public as including different places and different times. The mens rea is not
whether the underlying work is ©d but whether a performance will occur. [What
distinguishes that from a copy shop?]
A: the copy
shop has set up technology that allows the user to make a copy, who is causally
responsible. But the webcam creates a transmission that would otherwise not
proportion of infringing uses can’t be the distinction; even if it’s substantially
certain that infringement will occur, you can’t be sure which or when. It has
to be about whether the overall purpose is ok. The purpose of the copy shop is
to allow copying of things, some of which will infringe; the purpose of Aereo
is to retransmit network signals, where it’s not a matter of uncertainty about
what will be retransmitted.
A: important to
consider equities between user and technological provider. The user in the copy
shop has to take a book to the copy shop and make a choice about what to copy,
whereas the user of Aereo will not be liable on their own because they aren’t
making a public performance. [This seems to assume the conclusion: there must
be liability for one party, as opposed to the conduct being legal in its
entirety.] Maybe we’re going back to Breyer’s functional analysis, but for
cases like Polyvore it may provide a more satisfying analysis.
Sag: is this
principle or policy?
A: trying to
start from first principles. The waiter who delivers the poison is less
causally responsible than the person who creates the poisonous drink. [I think
this embeds the conclusion of whether there was poison in the drink in the
analysis of who is responsible for delivering the poison. If it’s not public
performance/display, then it’s just a drink, not a poisonous drink.] Being a button
pusher is not the sine qua non of liability, though maybe indirect liability
gets to the question easier.
Mark A. Lemley and Mark P. McKenna, Trademark Spaces
There is a space on a product where TMs go—on the back of
a computer in the center, as a logo on clothing. But others are harder.
Stitching on pockets? Sometimes hard to tell whether something is a design or a
brand. Seabrook Farms: design elements around a name/logo are not marks. By
contrast, Georgi vodka case says that a big O can count as a TM and may be inherently
distinctive even though it circles a logo; maybe it’s part of the logo itself.
Other packaging elements can serve as middle elements, like color on packaging/multiple
colors on packaging. Finally, Tom Lee’s fabulous empirical studies: when you
put something in the “TM space” on a box of cookies, many people will perceive
it as a TM even with a generic term.
Courts and the TTAB are evaluating branding at step zero
for both products and packaging, without taking evidence, often without knowing
they’re doing it, based largely on intuitive sense of whether consumers are
likely to view an image as serving a tm function. Much of that intuition stems
from TM spaces. TM owners can create new spaces, conditioning user response,
but doing so requires more than secondary meaning for one mark—Louboutin’s red
soles don’t mean that any color sole automatically functions as a TM. TTAB and
courts should make findings on TM spaces; we should require secondary meaning
as a default outside an established TM space. Even if you are in a TM space,
you still need to show distinctiveness. Cts/TTAB should make broader findings
about whether a space is a TM space generally. W/understanding of inherent
distinctiveness as a place, makes it easier to require secondary meaning as a
default in places outside the TM spaces. But putting generic word in TM space
shouldn’t generate ownership of that generic word. If in a TM space, do normal
Two Pesos analysis; if not in a TM space, secondary meaning required regardless
of whether inherent distinctiveness might be possible in another space.
Linford: what do you do about the fact that TM owners are
now willing to slap logos everywhere, e.g. on race cars. [I will note that’s
not everywhere generally; it is everywhere on race cars, and that might matter!
Betsy Rosenblatt has a similar response: that may ID what is actually an ad space
not an ordinary product, and racecars and sports jerseys are now equivalent to
McKenna: some TMs may only work as marks because their
placement is rare. [tragedy of the commons!]
Jeanne Fromer: Brunetti’s lawyer had a discussion of
Lemley: Lululemon registration case is a really interesting
example: moving a logo outside of a Tm space so it looks like a stylized thing
on a jacket may not move the meaning.
RT: (1) Booking may make your recommendations of limits
(when FUDGE COVERED COOKIES is in the TM space it’s still generic) impossible
because restrictions will always be on scope of rights not existence of the
mark in the first place. (2) Interaction with Grace McLaughlin’s article on marks in the TM space that fail to function nonetheless. (3) Descriptiveness refusals
in ITU cases— Eastman Kodak v. Bell & Howell versus the 1/64th Mattel
case—provide useful analysis.
Lemley: on (1), yeah that’s a problem; maybe everything is
malleable and TM is whatever you say loudly enough is a mark.
Rothman, Navigating the Identity Thicket: Trademark’s Lost Theory of
Personality, The Right of Publicity, and Preemption
The interface b/t TM and ROP is a mess—Joseph Abboud case isn’t just a
fair use case; court avoids the question of whether he could transfer his ROP
by holding he didn’t, and the court found no false endorsement, but those
questions don’t go away b/c one court dodges them. People are starting to claim
these kinds of things more commonly. Not always owned by same P—high profile cases
involving chefs who purportedly transferred both TM and ROP to another and then
transferee claims that they can’t use their own names on their own restaurants.
Happened with a wood-fired pellet company too.
Hubert Hansen IP Trust v. Coca-Cola led to $10 million jury award, more
than what Michael Jordan got; CC owned Hansen’s Soda founded by Hubert Hansen
(d. 1951 when postmortem rights didn’t exist). Grandkids sued for ROP violation. Even assuming that Hansen’s family succeeded
to postmortem rights, how to reconcile that with CC’s ownership of the TMs? CC’s
ability to use photos of him, references to his life story? If CC owns all the
IP except the ROP, what does that mean for the successors? Can they start their
own juice business and use his name in advertising?
Conflict preemption as a possibility. Claim: federal TM has as one
objective protecting autonomy-based and dignity-based personality interests,
along with the others that are more commonly referenced. (1) Natural/sacred
right to one’s name; (2) limits on using another person’s identity; (3) limits
on transferability of personal marks; (4) limits on abandonment. These doctrines
are largely intact from the 1800s even if we have forgotten why they exist.
Implications: we should further restrict alienability of personal marks
(distinguish de jure and de facto marks—can’t always separate identity from
person—Ford Motor Co.—we might initially have thought Ford integral to the
corporation, but now we know it’s distinct); we should continue robust latitude
to use one’s own name/identity after transfers of marks; shores up TM’s
negative spaces by permitting confusion when we tell the truth/provide
information about self; and finally explains TM’s expansionist impulses and
provides a means of limiting them when only corporations and not natural persons
Implications: Hansen: CC has the rights to do what it did, but Hansen heirs
have rights to use name in juice though maybe not as marks.
Abboud: if didn’t transfer ROP, has a lot of flexibility to do what he
Two bad options: ROP swallows up TM to become mutant TM law, or if we don’t
have that, we have TM law that vastly overlooks and shuts down legitimate
personality-based interests often asserted as ROP but also are adequately
protected under TM properly understood.
McKenna: historically names wouldn’t have been (technical) TMs; this would
have been unfair competition law. Limits on the remedies in these cases would
have been true in any unfair competition case. All of unfair competition got
assimilated into TM; all the remedial modesty went away. Is there anything
special about names or is it just a consequence of the collapse of all unfair
competition law into TM? It’s fine to say that names should be pulled back out
for specific reasons, but why it happened may matter.
A: paper covers differences in context of personal marks.
Victoria Schwartz: Negative v. positive rights: ability to tell one’s own
story versus the ability to block others from telling a related story—may deserve
different treatment. Name versus other aspects of identity may also change the
analysis. Could CC really tell Hansen’s life story in its ads? Is that the same
as using his name?
RT: consider literature on ideological drift/preservation through transformation—right
now you seem to be resting a lot on “we used to do it this way.” But it would
be actively shocking if 1800s judges understood how celebrity and market
dynamics worked now. These doctrines crumbled at the edges for reasons that
seemed good to their proponents; I think you should confront more directly that
it’s not obvious why 1800s cases would be right now.
A: Thinks there is a normative case for separating out personality
interests and will defend that directly.
Felix Wu: names v. other things?
A: interesting changes in this over time/different cases—signature/likeness
may not transfer with transfer of name because more connected to moral
from Blogger https://ift.tt/2WX5p23