Sign in

Blake E. Reid

@chup.blakereid.org
5.5K followers 1.1K following 1.1K posts

Simple country tech law professor, multidisciplinary dilettante, “idea generator.” Crocs, fizzy water, ska. Someone is wrong on the Internet and it’s probably me. No legal or good advice; opinions my own, bad. Supposed to be on posting hiatus.

PostsRepliesMedia
Blake E. Reid @chup.blakereid.org · 06/10/2026
This bit from Reno’s a pretty solid pointer to when things went south
The Act includes seven Titles, six of which are the product of extensive committee hearings and the subject of discussion in Reports prepared by Committees of the Senate and the House of Representatives. By contrast, Title V-known as the "Communications Decency Act of 1996" (CDA)-contains provisions that were either added in executive committee after the hearings were concluded or as amendments offered during floor debate on the legislation. An amendment offered in the Senate was the source of the two statutory provisions challenged in this case
010
Blake E. Reid @chup.blakereid.org · 18/09/2026
The OP is good example of writing a heading and an article to avoid misinforming people. It's the "US military using AI" and "the analyst queried a chatbot," not "rogue AI almost starts World War III." It prompts the reader to wonder why the military and analyst used this tech and why it failed.
Exclusive: US military had close call after using AI for false intelligence report, sources say
230
Blake E. Reid @chup.blakereid.org · 17/09/2026
I'm not saying it won't happen here, but in this case the statements are not just contestably relevant evidence. They literally encompass the whole thing that the cause of action prohibits, on purpose!
Carr’s threats presented a textbook unlawful jawbone, controlling for the
variables that often bedevil the legal assessment of government pressure on
speech intermediaries.14 Carr’s movie-villain threats, described even by
Republican Senator (and Carr ally) Ted Cruz as “right out of Goodfellas,”15
so flagrantly threatened ruinous consequences without any plausible legal
basis16 that their unlawfulness was self-evident to many legal scholars.17
By holding unlawfulness constant, Carr’s threats neatly highlight the
willingness of speech intermediaries to resist unlawful jawboning18
—a
public good19 this Essay calls jawbreaking.20
250
Blake E. Reid @chup.blakereid.org · 02/09/2026
Substance aside, pretty extraordinary to see the DOJ attack the credibility of the Register of Copyrights in a public filing like this (from the U.S. OpenAI copyright MDL).
17 The Kadrey court did not cite any case supporting such a theory. But the Register of Copyrights, who is currently challenging her removal, appeared to endorse a similar theory in a report. See U.S. COPYRIGHT OFFICE, COPYRIGHT AND ARTIFICIAL INTELLIGENCE PART 3:
GENERATIVE AI TRAINING 64 (May 2025), https://perma.cc/Q2TQ-Y37B. Her understanding does not warrant deference. Loper Bright Enters. v. Raimondo, 603 U.S. 369, 412 (2024). And the Register's threadbare reasoning ignored all the caselaw emphasizing the required use-by-use analysis and discussing what sort of "effect" upon the potential market is cognizable. Id. Although
140
Blake E. Reid @chup.blakereid.org · 02/07/2026
Via analysis supposedly on the other site (which I can't see directly, but attributed to @jacklgoldsmith.bsky.social), there was a brief but very illuminating exchange that I missed between Chief Justice Roberts, Justice Sotomayor and SG Sauer at the Slaughter oral arguments about Perlmutter.
CHIEF JUSTICE ROBERTS: I think there
are a lot of agencies in the federal government
where it's hard to parse whether it's an
executive function they're engaged in or a
legislative function. We obviously have the
Perlmutter case holding, where you do -- deal
with the Library of Congress, which half of
it's a library, half of it's things like the
copyright. What are we supposed to do with
that if you're correct?
GENERAL SAUER: Well, Mr. Chief
Justice, in Free Enterprise Fund, this Court I
think very aptly stated that the vast and
varied nature of the federal government is a
reason not to make general pronouncements onCHIEF JUSTICE ROBERTS: I think there
are a lot of agencies in the federal government
where it's hard to parse whether it's an
executive function they're engaged in or a
legislative function. We obviously have the
Perlmutter case holding, where you do -- deal
with the Library of Congress, which half of
it's a library, half of it's things like the
copyright. What are we supposed to do with
that if you're correct?
GENERAL SAUER: Well, Mr. Chief
Justice, in Free Enterprise Fund, this Court I
think very aptly stated that the vast and
varied nature of the federal government is a
reason not to make general pronouncements onJUSTICE SOTOMAYOR: -- instead of the
removal power?
GENERAL SAUER: Well, for example,
consider an agency like the FTC, which is
before the Court. Virtually all of its
powers -- I can't think of a power that it
exercises that is not executive, so there's
nothing to --
140
Blake E. Reid @chup.blakereid.org · 21/05/2026
Because *of course*, a late-breaking update: the House Appropriations Committee approved the House's FY27 leg. branch appropriations bill, which includes this wacky manager's amendment that gives appointment and removal authority over the Librarian to a big committee of House and Senate leadership.
S e c .
_ Notwithstanding section 2 of the Librarian of Congress Succession Modernization Act
of 2015 (2 U.S.C. 136-1) and section 301 of title 44, United States Code, the Librarian of
Congress and the Director of the Government Publishing Office shall be appointed, and may be
removed, without regard to political affiliation and solely on the basis of fitness to perform the
duties of the office, upon a majority vote of a congressional commission consisting of the
Speaker of the House of Representatives, the majority leader of the Senate, the minority leaders
of the House of Representatives and Senate, the chair and ranking minority member of the
Committee on Appropriations of the House of Representatives, the chairman and the ranking
minority member of the Committee on Appropriations of the Senate, the chair and ranking
minority member of the Committee on House Administration of the House of Representatives,
and the chairman and ranking minority member of the Committee on Rules and Administration
of the Senate.
121
Blake E. Reid @chup.blakereid.org · 29/03/2026
Gonna have to update the chart I guess
Flow chart:
In red, "I'm having an Internet problem" flows to a decision point: "Is it copyright?" "Yes" flows to "Copyright takedown" in green. "No" flows to another decision point labeled "Yes it is," and the only answer, "Okay," flows to "Copyright takedown"
081
Blake E. Reid @chup.blakereid.org · 06/07/2025
Grateful to @indefenseofska.bsky.social for so many years of good listens and reconnecting me with a scene that I’d lost. And very proud to have dropped what I *think* is the only cite (so far!) to the book in a law review article. (Still has to survive the editors but I will fight for it.)
Of course, any reasonably self-conscious telecom law scholar must admit that the field has fallen out of favor since the late 1990s499 when internet exceptionalism began to take root.500 More problematically, Justice Kagan’s…499. Defenders of ska music know the feeling. Cf. Aaron Carnes, IN DEFENSE OF SKA (2021).
253
Blake E. Reid @chup.blakereid.org · 30/06/2025
The first reg implicates a mess of questions around the "generally applicable law" carveout. The second, likewise, around the "facilitate the development or operation of" carveout. 4/
‘‘(II) a generally applicable law or
regulation, such as a law or regulation per-
taining to unfair or deceptive acts or prac-
tices, child online safety, child sexual abuse
material, rights of publicity, protection of a
person’s name, image, voice, or likeness
and any necessary documentation for en-
forcement, or a body of common law, that
may address, without undue or dispropor-
tionate burden, artificial intelligence mod-
els, artificial intelligence systems, or auto-
mated decision systems to reasonably effec-
tuate the broader underlying purposes of
the law or regulation; and‘‘(I) remove legal impediments to, or
facilitate the deployment or operation of,
an artificial intelligence model, artificial in-
telligence system, or automated decision
system; or
170
Blake E. Reid @chup.blakereid.org · 30/06/2025
Second, there's the regulation of third parties that provide age verification systems, which arguably *compels* the use of an some kind of identification system (all of the possibilities of which are arguably themselves ADS) and imposes a nominal privacy regulation on those parties 3/
(b)  A commercial entity that performs the age verification
 	required by Subsection (a) or a third party that performs the age
 	verification required by Subsection (a) may not retain any
 	identifying information of the individual.
160
Blake E. Reid @chup.blakereid.org · 30/06/2025
First, there's the regulation of "knowing and intentional publication on an Internet website" which at any scale is going to involve something at least in the ballpark of what the budget bill calls an "automated decisionmaking system" 2/
	       Sec. 129B.002.  PUBLICATION OF MATERIAL HARMFUL TO MINORS.
 	(a)  A commercial entity that knowingly and intentionally publishes
 	or distributes material on an Internet website, including a social
 	media platform, more than one-third of which is sexual material
 	harmful to minors, shall use reasonable age verification methods as
 	described by Section 129B.003 to verify that an individual
 	attempting to access the material is 18 years of age or older.‘‘(E) AUTOMATED DECISION SYSTEM.—
The term ‘automated decision system’ means
any computational process derived from ma-
chine learning, statistical modeling, data ana-
lytics, or artificial intelligence that issues a sim-
plified output, including a score, classification,
or recommendation, to materially influence or
replace human decision making.’’; and
160
Blake E. Reid @chup.blakereid.org · 30/06/2025
Tired: leaving the Internet largely unregulated for three decades because judges aren’t sure what the word “publisher” means Wired: replacing one word with big word salad that will now be litigated in every state law case nominally involving a computer and a decision
“(II) a generally applicable law or
regulation, such as a law or regulation per-
taining to unfair or deceptive acts or prac-
tices, child online safety, child sexual abuse
material, rights of publicity, protection of a
person’s name, image, voice, or likeness
and any necessary documentation for en-
forcement, or a body of common law, that
may address, without undue or dispropor-
tionate burden, artificial intelligence mod-
els, artificial intelligence systems, or auto-
mated decision systems to reasonably effec-
tuate the broader underlying purposes of
the law or regulation; and
15924
Blake E. Reid @chup.blakereid.org · 27/06/2025
Cosign @kendraserra.bsky.social's suggestion to read Kagan's dissent first. It cuts right to the chase: Thomas' opinion is openly results-oriented, but sloppily tries to cover that up with a mess of self-contradictory reasoning from which everyone else will have to try and divine a coherent rule.
The majority’s opinion concluding to the contrary is, to be
frank, confused. The opinion, to start with, is at war with
itself. The majority is not shy about why it has adopted these
special-for-the-occasion, difficult-to-decipher rules. Itthinks they are needed to get to what it considers the right
result: giving Texas permission to enforce its statute.
150
Blake E. Reid @chup.blakereid.org · 27/06/2025
As a bonus from FSC v. Paxton, don't sleep on fn. 4, where Thomas can't help but continue his extra-textual quest to reinterpret Section 230 to mean something different from what it says on the tin. (At least he cited it correctly.)
4 Elsewhere in the CDA, Congress recognized that content filtering was
still an emerging technology and that companies attempting to use it
faced serious risks. A year before the CDA’s enactment, a New York
court had held that an online service provider could be held liable as a
publisher for defamatory posts by third-party users because the provider
had “held itself out as” “a family oriented computer network” that
screened out inappropriate content. Stratton Oakmont, Inc. v. Prodigy
Servs. Co., 1995 WL 323710, *2 (Sup. Ct. N. Y., May 24, 1995). In re-
sponse, the CDA added a new §230 to the Communications Act of 1934.
§509, 110 Stat. 137–139 (codified as amended at 47 U. S. C. §230). Sec-
tion 230 provides that computer service providers (1) shall not “be
treated as the publisher or speaker of any information provided by” a
third party, and (2) shall not “be held liable” for good-faith actions to re-
strict access to material that they consider to be “obscene, lewd, lascivi-
ous, filthy, excessively violent, harassing, or otherwise objectionable,” or
to enable others (such as users) to restrict access to such material.
§230(c). Congress thereby aimed to spur the development and use of fil-
tering technology so that parents could prevent their children from ac-
cessing sexually explicit content online. See §509, 110 Stat. 137 (title)
(“Online Family Empowerment”); 47 U. S. C. §230 (title) (“Protection for
private blocking and screening of offensive material”); §230(c) (title)
(“Protection for ‘Good Samaritan’ blocking and screening of offensive ma-
terial”); 141 Cong. Rec. 22045 (1995) (remarks of Rep. Cox) (“We want to
encourage” computer service providers “to help us control . . . what our
children see” using filtering “technology” that “is very quickly becoming
available”).
171
Blake E. Reid @chup.blakereid.org · 27/06/2025
I also don’t think “leaving the Copyright Office without an active boss” is an accurate characterization of what Perlmutter’s lawyers argue will cause dysfunction. They’ve been quite explicit that she “remains the Register of Copyrights.”
2. Plaintiff remains Register of Copyrights and is therefore required to fulfill
her statutory duties.
Beyond the loss of her “statutory right to function,” absent preliminary relief Plaintiff will
also be prevented from performing the legislative functions that continue to be required of her.
Ms. Perlmutter remains the Register of Copyrights according to the Library of Congress’s
leadership and staff; she has not received any communication from Library of Congress leadership
that could lawfully terminate her employment. See Perlmutter Decl. ¶ 13; see also Leadership,
U.S. Copyright Off., https://www.copyright.gov/about/leadership (identifying Ms. Perlmutter as
Register as of June 10, 2025). Ms. Perlmutter therefore is required by law to fulfill her statutory
obligations while this lawsuit is pending, unless she is removed from office by a duly appointed
Librarian.
000
Blake E. Reid @chup.blakereid.org · 25/06/2025
Judge Chhabria knew that his opinion in Kadrey v. Meta would be publicly consumed and misunderstood, so he wrote a GIANT CAVEAT ABOUT THE LIMITED SIGNIFICANCE OF THE HOLDING THAT EVERYONE REALLY SHOULD READ BEFORE ISSUING HOT TAKES. It's very short and I'm not even going to summarize it. Read it!
The upshot is that in many circumstances it will be illegal to copy copyright-protected
works to train generative AI models without permission. Which means that the companies, to
avoid liability for copyright infringement, will generally need to pay copyright holders for the
right to use their materials.
But that brings us to this particular case. The above discussion is based in significant part
on this Court’s general understanding of generative AI models and their capabilities. Courts can’t
decide cases based on general understandings. They must decide cases based on the evidence
presented by the parties.Given the state of the record, the Court has no choice but to grant summary judgment to
Meta on the plaintiffs’ claim that the company violated copyright law by training its models with
their books. But in the grand scheme of things, the consequences of this ruling are limited. This
is not a class action, so the ruling only affects the rights of these thirteen authors—not the
countless others whose works Meta used to train its models. And, as should now be clear, this
ruling does not stand for the proposition that Meta’s use of copyrighted materials to train its
language models is lawful. It stands only for the proposition that these plaintiffs made the wrong
arguments and failed to develop a record in support of the right one.
35316
Blake E. Reid @chup.blakereid.org · 25/06/2025
...and just in case you thought Judge Alsup's framework for generative AI and copyright was going to uncontroversially become the law of the land, here's Judge Chhabria coming off the top rope with a sharp critique and a totally different approach
Speaking of which, in a recent rulingon this topic,Judge Alsup focused heavily on the transformative nature of generative AI while brushing aside concerns about the harm it can inflict on the market for the works it gets trained on. Such harm would be no different, he reasoned, than the harm caused by using the works for “training schoolchildren to write well,”which could “result in an explosion of competing works.” Order on Fair Use at 28, Bartz v. AnthropicPBC, No. 24-cv-5417 (N.D. Cal. June 23, 2025), Dkt. No. 231. According to Judge Alsup, this “is not the kind of competitive or creative displacement that concerns the Copyright Act.” Id.But when it comes to market effects, using books to teach children to write is not remotely like using books to create a product that a single individual could employ to generate countless competing works with a miniscule fraction of the time and creativity it would otherwise take.This inapt analogy is not a basis for blowing off the most important factor in the fair use analysis.
119122
Blake E. Reid @chup.blakereid.org · 24/06/2025
One last thing: I’ve said that these technologies often look like they are designed by engineers who’ve locked their copyright lawyers in the broom closet. I’ve been focused a lot on the user interfaces. But where this really materialized in this case was in the data set assembly.
Anthropic thereby pirated over seven million copies of books, including copies of at least two works at
issue for each Author.'
As Anthropic trained successive LLMs, it became convinced that using books was the
most cost-effective means to achieve a world-class LLM. During this time, however,
Anthropic became "not so gung ho about" training on pirated books "for legal reasons"
2235
Blake E. Reid @chup.blakereid.org · 20/06/2025
To be specific I’m not sure the info asymmetry is the same on either side or that the same kind of transparency is plausible
There is an inescapable information asymmetry between users and search engines. No one setting out on a journey of en- lightenment knows what lies at the end of the road—if she did, there would be no need of the journey. The user knows more about what she wants, whether it be free online calculus practice questions] or (brinty spiers toplessl, but the search engine knows far more about whether anyone has put practice questionsTransparency is a crucial aspect of loyalty. On one level,
proper disclosures can defuse almost any deception. But on a
deeper level, transparency is also profoundly helpful in ena-
bling users to understand what it is they are getting from a
search engine and how to use it effectively.
120
Blake E. Reid @chup.blakereid.org · 17/06/2025
It's unfortunate that Kevin and @caseynewton.bsky.social obscure what are directionally useful and journalistically reasonable descriptive points about AI's proliferation and social salience with anecdotal, performative provocations, half-baked policy prescriptions, and strawman-burning of critiques
ROOSE A.I. has essentially replaced Google for me for basic questions: What setting do I put this toaster oven on to make a turkey melt? How do I stop weeds from growing on my patio? I use it for interior decorating — I’ll upload a photo of a room in my house and say, “Give this room a glow-up, tell me what furniture to buy and how to arrange it and generate the ‘after’ picture.” A friend of mine just told me that they now talk to ChatGPT voice mode on their commute in their car — instead of listening to a podcast, they’ll just open it up and say, “Teach me something about modern art,” or whatever.

NEWTON That’s a terrible threat to our business. What are we doing about this?

ROOSE I guess it’s time to pivot to modern art. Another person I know just started using ChatGPT as her therapist after her regular human therapist doubled her rates.If you’re the first lawyer who cites a nonexistent case because of ChatGPT, that’s on ChatGPT. If you’re the 100th, that’s on you.Sometimes when I hear people arguing about how A.I. systems are stupid and useless, it’s almost as if you had an antinuclear movement that didn’t admit fission was real — like, looking at a mushroom cloud over Los Alamos, and saying, “They’re just raising money, this is all hype.” Instead of, “Oh, my God, this thing could blow up the world.”

NEWTON Yeah, I think so much A.I. denialism comes off as a kind of wishful thinking — which, again, I’m sympathetic to, because in a lot of ways it would be easier if all this stuff was fake and was going to fall into the ocean the way that cryptocurrency did after its 2021 peak. But as journalists, the more we talk to people, the less likely we think that is.
46010
Blake E. Reid @chup.blakereid.org · 06/06/2025
"No coding needed" tools long predate generative AI and the perennial problem with them is that maintenance and updates inevitably create dependencies. (The long term mechanisms for maintaining and updating a codebase are as or more important than the code for a software project of any complexity.)
For personal productivity tools, workflow automation, and small community projects, you can actually build what you need, no coding skills needed. The tools exist, they’re accessible, and they work. Sometimes the best way to escape someone else’s control is to just stop asking, and just take charge.
070
Blake E. Reid @chup.blakereid.org · 06/06/2025
A (truly) friendly suggestion: next time you want to make a significant change, try doing it with another tool (or do it manually, or let a coder friend take a look and give you an estimate for what they'd charge). If the experience is along the lines of what you've described here...
Then, at the suggestion of a friend, I tried to import the project into Firebase Studio, owned by Google, and built on top of Firebase. As with Bolt, I first tried to pull in my GitHub repo, which worked… except, for reasons I don’t understand, the AI features then all seemed disabled. After messing around with it for a bit and getting nowhere, the same friend who recommended I try Firebase Studio said “it might just be faster to ask Firebase Studio to recreate the project from scratch.” I gave it the same prompt and… it created “Momentum Flow,” with its own interpretation of the prompt.
200
Blake E. Reid @chup.blakereid.org · 06/06/2025
Glad to see @waldo.net making the observation that should be beat-you-over-the-head obvious to any CS major who's taken a software engineering course, which is that problem assessment and refinement and tool selection are at the threshold of any major software project.
“AI is absolutely the wrong tool for this,” said Waldo Jaquith, a former Obama appointee who oversaw IT contracting at the Treasury Department. “AI gives convincing looking answers that are frequently wrong. There needs to be humans whose job it is to do this work.”
215217
Blake E. Reid @chup.blakereid.org · 05/06/2025
Lastly, Judge Hinkle is already tired of this case. He won't compel detailed discovery from all of NetChoice's members because, to summarize, the state law as drafted is a total dumpster fire and he doesn't need to know all the gory details of every internet company to figure that out. 12/
The Supreme Court’s discussion of the plaintiffs’ facial claims and the
breadth of the inquiry those claims will require has understandably triggered every
good litigator’s best, or perhaps worst, instinct: to turn over every rock, to catalog
every fact, to explore every nuance. But this litigation need not be nearly that
complicated. Some of these challenged provisions, as applied to some of the
plaintiffs’ members, are almost surely unconstitutional; at oral argument, the
defendants came close to acknowledging this. With “social media platform” now
properly construed and the required facial-challenge methodology now settled by
the Supreme Court, the plaintiffs’ facial challenge to SB 7072, and perhaps even toits various provisions viewed in isolation, is likely to fail—and the disputed
discovery, if allowed, would almost surely make no difference. Discovery that
imposes a significant burden and is almost sure to make no difference is ordinarily
beyond the properly defined scope of discovery; it fails the proportionality test. If,
as the litigation progresses, it turns out this is incorrect—if it turns out the facial
challenge is more substantial than this order suggests and that additional discovery
might enhance the defendants’ ability to fully present their position—the discovery
issue can be revisited.
1202
Blake E. Reid @chup.blakereid.org · 04/06/2025
Also not surprising to see a more detailed treatment of vagueness here, given how unclear it is how these statutes actually work. (Cf. @mmasnick.bsky.social's colorfully titled piece on the Texas statute. www.techdirt.com/2022/05/12/j...) 11/
 no way to determine what the provision really requires.
The defendants have not attempted to explain what these provisions really
mean or how they would be applied. Nor have the defendants offered any theory
under which a state can preclude this kind of curating without violating the First
Amendment.
1121
Blake E. Reid @chup.blakereid.org · 04/06/2025
It *doesn't* look like the court will simply rest on its previous 230 preemption holding, but rather will take on some more searching review of the question, presumably given the SCOTUS demand for details on the 1A issues. 9/
In sum, Madonna is still the law of the circuit. In due course, count four will
require a searching, provision-by-provision analysis.
1120
Blake E. Reid @chup.blakereid.org · 04/06/2025
More importantly, the court tees up the possibility that it will knock out at least some of the issues in the case with Section. 230 rather than the First Amendment, which I have long speculated might happen. (Indeed, this court did just that in the first round of litigation!) 8/
C. § 230 merits
Count four of the amended complaint alleges § 230 preempts specific parts,
not all, of Florida Statutes §§ 106.072 and § 501.2041. The claim is labeled neither
facial nor as-applied; it is simply a preemption claim.Second, however: the proceedings also underscore that
Section 230 may preempt both laws to some degree—a
possibility that arose at the NetChoice oral arguments.169 As to the
Florida law, the district court in Moody quietly concluded that
Section 230 preempted the parts of the Florida law “that purport
to impose liability for other decisions to remove or restrict access
to content.”170 The Eleventh Circuit did not address or disturbthe district court’s conclusions about preemption of the carriage
provisions.171
As for the Texas law, the district court offhandedly
acknowledged but did not address the platforms’ Section 230
preemption arguments.172 Before the Fifth Circuit, the platforms
briefly reasserted that Section 230 preempted H.B. 20, but
declared that there was “no need for [the Fifth Circuit] to address
[Section 230 preemption].”173 The Fifth Circuit castigated the
platforms for addressing preemption so briefly, concluding that
the treatment was “insufficient to adequately brief [the] claim”
and that the platforms had “forfeited their preemption
argument.”174 Yet even if the Fifth Circuit’s ruling on preemption
in Paxton itself holds following the Supreme Court’s ruling, the
possibility remains that platforms will be able to reassert Section
230 preemption in future facial or as-applied challenges.
Finally, the proceedings underscore that Florida and
Texas did not undertake aggressive efforts to pay down Section
230’s legislative debt in ignorance of the possibility that Section
230 might preempt their laws even if they shot the moon on the
First Amendment with the Supreme Court. Both Texas and
Florida aggressively urged the courts from the early stages of
both Moody and Paxton to narrow the scope of Section 230.175
And while the Supreme Court’s grant of certiorari in the
NetChoice cases nominally is limited to questions about the First
Amendment,176 both Florida and Texas have sought throughoutthe litigation to sneak Section 230 back into the cases using the
First Amendment as a Trojan horse.177 Without regard to the
contestable substance of the arguments, it is underappreciated
how aggressively the states have urged the Court to revisit its
decision to punt on the scope of Section 230 in Gonzalez178 with a
vehicle potentially more likely to appeal to the Court’s
conservative justices.179
1110
Blake E. Reid @chup.blakereid.org · 04/06/2025
Anyway, I'm not sure how novel this is, but it's pretty interesting to see the court discuss the use of Section 1983 as a claim against state laws that trod on Section 230. 7/
A hypothetical may help explain just how far the defendants’ position strays
from the proper analysis. Suppose a state passed a statute saying a social-media
provider could be held liable for damages based on any defamatory material posted
on its site—precisely the result § 230 was enacted to prohibit—and that each
posting of defamatory material would result in a substantial fine levied against the
provider, regardless of whether the provider was even aware of the defamatory
material. Any assertion that a provider who hosted millions of postings each day
could not seek prospective relief under § 1983 or equitable principles would be, in
the language of Moms for Liberty, borderline frivolous. Or perhaps not just
borderline.
1141
Blake E. Reid @chup.blakereid.org · 04/06/2025
The court makes a sly little dig at Justice Thomas by pointing out at that he raised standing as an issue when the case was before SCOTUS, and implies that the rest of the Justices tacitly blessed standing by ignoring him. LOL. 6/
The Supreme Court, too, addressed the merits in substantial respects,
explicitly providing guidance to the courts on remand, without questioning the
plaintiffs’ standing. The Court could not have overlooked the issue; it was raised ina separate opinion. See Moody v. NetChoice, 603 U.S. at 760 n.2 (Thomas, J.,
concurring in the judgment). The defendants say the Court’s failure to challenge
standing was just a “drive-by” ruling that should be ignored. Perhaps so. But it was
a drive-by ruling in this very case.
1221
Blake E. Reid @chup.blakereid.org · 04/06/2025
Anyway, common carriage sloganeering doesn't work on the internet, because we have lots of different definitional buckets that work in counterintuitive ways. Here, the Florida statute applies beyond traditional social media because the drafters sloppily based their definitions on Section 230. 4/
The controlling definition is of course the definition that was enacted into
law, not the description the Senate sponsor provided in support of a proposed
amendment that failed. The inescapable fact is that the adopted language derived
from § 230(f)(2). When context and all appropriate considerations are factored in,
the adopted definition still means what it says.
In sum, both § 230(f)(2) and Florida Statutes § 501.2041(1)(g) apply not just
to traditional social-media providers but also to a wide variety of other applications
that provide multiple users access to or over the internet.
1130
Blake E. Reid @chup.blakereid.org · 04/06/2025
Aside: I could not agree more with @ericgoldman.bsky.social on the nebulousness of the term "ISP," which is one reason my coauthors and I spend a whole chapter in our textbook on classification problems! 3/
It turns out that for the internet, the substantive policy question of access has
effectively collapsed into the more formalistic legal question of classification: What
is this thing called “the internet”? This is why we’ve pulled out internet access into
a separate chapter: to use it as the pedagogical vehicle with which to explore the
important concept of classification.
1100
Blake E. Reid @chup.blakereid.org · 04/06/2025
First, I'm thrilled that the court frames the case first as a classification problem—a key telecom concept, which further underscores my argument that NetChoice is a telecom case! 2/
At the outset, a word is in order about terminology. This order uses the term
“traditional social-media provider” to refer to what most people would probably
understand that term to mean—so Facebook, Instagram, YouTube, X, and dozens
of smaller but similar providers. The distinguishing characteristic is perhaps this:
the primary function of a traditional social-media provider, or at least a primary
function, is to receive content from users and in turn to make the content available
to other users, usually allowing those other users to respond. See, e.g., Moody v.
NetChoice, 603 U.S. at 719. This is hardly a precise definition, but none is needed;
the term is used only for purposes of this order. The term “traditional social-media
provider,” as used in this order, is intentionally not a description of the providers
who are covered by the challenged statutes.
The challenged statutes, in contrast, use a different term, “social media
platform.” See Fla. Stat. § 501.2041(1)(g). Regardless of which side is correct
about the proper construction of that term’s statutory definition, it is clear that notall traditional social-media platforms are covered; to be covered, an entity must
meet minimum revenue or participant requirements. See id. § 501.2041(1)(g)4.
There is no significance to this order’s use of “provider” instead of “platform.” It is
just an effort to further distinguish two different sets of entities: traditional social-
media providers, on the one hand, and the social media platforms covered by the
statute, on the other hand.This Article explains these significant implications of NetChoice’s core holding
by framing NetChoice as a new entry in the telecom First Amendment canon,
sitting atop a foundation of underexplored and misunderstood jurisprudence
dating back nearly a century.
14
2130
Blake E. Reid @chup.blakereid.org · 03/06/2025
See the parallels to the arguments about curated collections (and Justice Barrett’s AI-specific concurrence) in NetChoice v. Moody, encapsulated in the nastiest footnote I have ever written (H/T @margotkaminski.bsky.social)
See Balkin, supra note 5. See discussion supra, Part II.B.ii. The threshold salience of speech also
features significantly in Justice Alito’s concurrence in the judgment. See NetChoice 703 U.S. at
781–82 (Alito, J., concurring in the judgment) (arguing that “not all compilers express a
message of their own” and therefore that “not all compilations are protected by the First
Amendment”). The focus in NetChoice’s dicta on compilation raises an important connection
to the concept in copyright law. Cf. Feist Publications v. Rural Tel. Serv. Co., 499 U.S. 340, 1289(1991) (“The compilation author typically chooses which facts to include, in what order to
place them, and how to arrange the collected data so that they may be used effectively by
readers. These choices as to selection and arrangement, so long as they are made
independently by the compiler and entail a minimal degree of creativity, are sufficiently
original that Congress may protect such compilations through the copyright laws.”); Derek E.
Bambauer, Copyright = Speech, 65 EMORY L.J. 199, 200 (2015) (“if expression can be
copyrighted, . . . then that expression is speech that enjoys First Amendment protection.”). But
cf. NetChoice, 603 U.S. at 746 (Barrett, J., concurring) (“[W]hat about AI, which is rapidly
evolving? What if a platform’s owners hand the reins to an AI tool . . . ?”); Thaler v. Perlmutter,
687 F. Supp. 3d 140, 149 (D.D.C. 2023) (“The increased attenuation of human creativity from
the actual generation of the final work will prompt challenging questions regarding how
much human input is necessary to qualify the user of an AI system as an ‘author’ of a
generated work . . .”). Then again, cf. Toni M. Massaro, Helen Norton, & Margot Kaminski,
Siri-Ously 2.0: What Artificial Intelligence Reveals About the First Amendment, 101 MINN. L. REV.
2481 (2017) (“Current free speech theory and doctrine support the extension of free speech
rights to strong AI speakers”). But cf. Eugene Volokh, Mark A. Lemley, Peter Henderson,
Freedom of Speech and AI Output, 3 J. FREE SPEECH L. 651, 651 (2023) (“[C]urrent AI programs
are of course not people and do not themselves have constitutional rights, [but] their speech
may potentially be protected because of the rights of the programs’ creators.”). Perhaps the
expressive ontology of internet platforms belongs on the long list of problems that copyright
cannot solve. See Reid, What Copyright Can’t Do, supra note 385. Thanks to Margot Kaminski
for the insight.
120
Blake E. Reid @chup.blakereid.org · 23/05/2025
they're good deals Brent
Anthropic’s use of notorious digital piracy websites raised concerns for Alsup. In response, the company’s counsel, Joseph Richard Farris of Arnold & Porter Kaye Scholer LLP, argued the Supreme Court has been skeptical whether bad faith has any effect on the fair use analysis.

Alsup pushed back, saying “I have a hard time seeing that you can commit what is ordinarily a crime, but get exonerated because you end up using it for a transformative use.”

There could be a scenario where the court holds Anthropic should pay for the initial copies it acquired, Alsup said, suggesting the cost could be determined using Amazon book prices. But he noted the relatively low cost of legitimate alternatives, saying he’s found books for $1 at library sales.
2120
Blake E. Reid @chup.blakereid.org · 23/05/2025
Wowza. This is what I mean when I say that the AI-fair use questions are wildly unpredictable—this would be an outcome that makes everyone somewhat unhappy (reposted with alt-text)
“I’m inclined to say they did violate the Copyright Act but the subsequent uses were fair use,” Judge William Alsup said Thursday during a hearing in San Francisco. “That’s kind of the way I’m leaning right now,” he said, but concluded the 90-minute hearing by clarifying that his decision isn’t final. “Sometimes I say that and change my mind.”
5193
Blake E. Reid @chup.blakereid.org · 23/05/2025
I was really hoping for more details about what went down here but basically we just get "the Librarian was fired and then yada yada yada, the pre-publication AI report was released." Where is paragraph 6.5???
6. As Register of Copyrights, by statute, I act under the Librarian of Congress’s
supervision and direction. Since May 8, 2025, when President Trump fired Librarian of Congress
Carla D. Hayden, I have acted under the supervision and direction of Robert R. Newlen, who
replaced Dr. Hayden as acting Librarian in accordance with the Library of Congress’s regulations.
7. On May 9, 2025, the U.S. Copyright Office issued in pre-publication format, the
long-awaited Part 3 of Copyright and Artificial Intelligence, a report made pursuant to my statutory
responsibility to “[c]onduct studies . . . regarding copyright” and “[a]dvise Congress on national
and international issues relating to copyright.” 17 U.S.C. § 701(b)(1), (b)(4).
340
Blake E. Reid @chup.blakereid.org · 23/05/2025
I don't want to tell a professional editor that you buried the lede here but, uh, "Monkey see, monkey sue" 🤣🤣🤣
A monkey, an animal-rights organization and a primatologist walk into federal court to sue
for infringement of the monkey’s claimed copyright. What seems like the setup for a punchline is
really happening. It should not be happening. Under Cetacean Community v. Bush, 386 F.3d 1169
(9th Cir. 2004), dismissal of this action is required for lack of standing and failure to state a claim
upon which relief can be granted. Monkey see, monkey sue is not good law – at least not in the
Ninth Circuit.
RELEVANT FACT
The only pertinent fact in this case is that Plaintiff is a monkey suing for copyright
infringement. Plaintiff’s factual assertions regarding the creation of the famous Monkey Selfie
photograph are fundamentally erroneous, but they must be accepted as true for purposes of this
Motion to Dismiss. Ashcroft v. Iqbal, 556 U.S. 662, 678 (2009). (Regardless, the true story of how
this photograph came to be is available at http://www.djsphotography.co.uk/original_story.html.)
140
Blake E. Reid @chup.blakereid.org · 22/05/2025
Character's reply doubles down on the entitlement by suggesting that expressive intent isn't even necessary. That's an argument, I guess? (cc @alanrozenshtein.com) But there's not even an in-the-alternative argument that there *is* expressive intent at play. storage.courtlistener.com/recap/gov.us...
C. No “Expressive Intent” Requirement Applies To Pure Speech
Plaintiff’s (and Amici’s) primary legal argument for categorically excluding AI-
generated speech from First Amendment protection—that the speech lacks
“expressive intent,
” Doc. 85 at 6–8, Doc. 96 at 8–11—misreads a body of case law. As
the cases Plaintiff cites make clear, the “expressive intent” requirement is part of the
“expressive conduct test,” which applies only “when analyzing the scope of the First
Amendment protection beyond ‘pure speech.’”
140
Blake E. Reid @chup.blakereid.org · 22/05/2025
In response, the plaintiffs rake Character over the coals with a fairly obvious argument: how can companion bots be expressive when their proprietors lack intentionality w/r/t, control over, or even understanding of the outputs? (See also the shoutout to @emilymbender.bsky.social in the footnote.)
C.AI argues that Plaintiff’s claims “seek to impose liability for expressive
content and violate the rights of millions of C.AI users to engage in and receive
protected speech.” Character Technologies, Inc.’s Mot. to Dismiss (“MTD”) at 1.
C.AI asks the Court to radically expand First Amendment protections from
expressions of human volition to an unpredictable, non-determinative system
where humans can’t even examine many of the mathematical functions creating
outputs, let alone control them. FAC ¶ 141. The Court should decline this offer.
To assert a First Amendment defense regarding human expression, there
must first be human expression. Expression requires a intention to convey ideas
or meaning. Films, songs, and video games work this way. Humans design the
specific plots, characters, words, and actions that communicate a message. LLM-
based products like Character.AI do not work like this. They serve as “stochastic
parrots,” automatically generating human language without understanding the
meaning of that which they generate. ¶¶ 23-24.
1 Defendants ask the court toradically expand First Amendment protections, from works where humans curate
every aspect to an unpredictable, non-determinative system where humans can’t
even examine many of the mathematical functions creating outputs, let alone
control them. FAC ¶ 141.
No court has ever ruled that a speaker engages in First Amendment-
protected expression when they compile computer-outputs using metrics they can
neither understand nor explain.. Defendants try to have it both ways. They claim
C.AI’s outputs are speech because they contain combinations of words. But they
avoid attributing C.AI’s messages to Shazeer, De Frietas, or any other human
shaping the product. Speech cannot be protected without human intentionality
behind it.The First Amendment is foundational to American democracy, creating
an edifice to ensure that private voices are heard, their beliefs respected, and their
government accountable. The freedom was not intended to be a hollow vehicle to
disseminate any kind of expression, and the Court should decline to accept such
an expansive interpretation here.1 The term was coined by Professor Emily Bender, who co-authored a paper, "On the Dangers of
Stochastic Parrots: Can Language Models Be Too Big?”, FAC ¶¶ 23-24, which identified risks inherent
to large language models.
272
Blake E. Reid @chup.blakereid.org · 22/05/2025
Just to drive the expressiveness point home, Character's Motion to Dismiss takes almost entirely for granted that companion bots are expressive. This is about as deep as the analysis gets. storage.courtlistener.com/recap/gov.us...
C.AI is not a product—as long-running authority confirms. C.AI’s service
delivers expressive ideas and content to users, similar to traditional expressive media
such as video games.But the context of the expressive speech—
whether a conversation with an AI chatbot or an interaction with a video game
character—does not change the First Amendment analysis.
150
Blake E. Reid @chup.blakereid.org · 22/05/2025
Worth reading the Garcia-Character AI opinion, which is not an anti-AI tirade. Rather, much like Moody, Paxton, and TikTok (and cf. the fair use fights) it showcases an entitled strategy from a tech company expecting to win if it asserts the 1A without showing its expressiveness math.
Instead, Defendants rest their conclusion primarily on analogy. (Doc. 98 at 6–
8). Defendants analogize interactions with Character A.I. Characters to interactions
with NPCs (non-player characters) in video games and interactions with other
persons on social media sites—both of which have received First Amendment
protection. (Id. at 8); see Brown v. Ent. Merchs. Ass’n, 564 U.S. 786 (2011); Moody
v. NetChoice, LLC, 603 U.S. 707 (2024); see also Wilson v. Midway Games, Inc.,
198 F. Supp. 2d 167, 181 (D. Conn. 2002) (“While video games that are merely
digitized pinball machines are not protected speech, those that are analytically
indistinguishable from other protected media, such as motion pictures or books,
which convey information or evoke emotions by imagery, are protected under the
First Amendment.”); Watters v. TSR, Inc., 715 F. Supp. 819, 821 (W.D. Ky. 1989)
(“[First Amendment] protection extends to publications such as ‘Dungeons and
Dragons,’ whether they are disseminated for the purpose of informing the public or
merely for providing entertainment.”). Defendants however do not meaningfully
advance their analogies.By failing to advance their analogies, Defendants miss the operative question.
This Court’s decision as to the First Amendment protections Character A.I. receives,
if any, does not turn on whether Character A.I. is similar to other mediums that havereceived First Amendment protections; rather, the decision turns on how Character
A.I. is similar to the other mediums. See Brown, 564 U.S. at 790; Moody, 603 U.S.
at 729–30. For example, in holding that video games receive First Amendment
protection, the Supreme Court reasoned:
Like the protected books, plays, and movies that preceded them, video games
communicate ideas—and even social messages—through many familiar literary
devices (such as characters, dialogue, plot, and music) and through features
distinctive to the medium (such as the player’s interaction with the virtual world).
That suffices to confer First Amendment protection.
Brown, 564 U.S. at 790. Similarly, in recognizing that editorial functions of social
media sites receive First Amendment protections, the Supreme Court reasoned:
A private party’s collection of third-party content into a single speech product (the
operators’ “repertoire” of programming) is itself expressive, and intrusion into that
activity must be specially justified under the First Amendment. . . . [However,] a
First Amendment claim will not succeed when the entity objecting to hosting third-
party speech is not itself engaged in expression.
Moody, 603 U.S. at 729–30.
4266
Blake E. Reid @chup.blakereid.org · 22/05/2025
Reminder that this website has a second button that you can also click
Bluesky's "Post" buttonBluesky's "Cancel" button
2100
Blake E. Reid @chup.blakereid.org · 19/05/2025
May I refer you to Turner Broadcasting II, a case nominally about how geting to watch broadcast channels on your cable box implicates the First Amendment that has a dissent with about 20,000 words of analysis like this
In considering whether the data on drops and carriage refusals tend to demonstrate a threat to “the continued viability of local broadcast television,” Turner, 512 U.S. at ––––, 114 S.Ct. at 2470, I note that the defendants repeatedly characterize the stations under threat of drop—i.e., those likely to elect must-carry—as marginal. See, e.g., Noll Deposition at 17–18; id. at 60 (absence of must-carry would lead to dropping of stations of “relatively low audience rating”); id. at 221–22 (the less popular a station the more important is must-carry to its welfare); id. at 307–08 (nothing in his declaration or testimony says cable systems will drop the “more popular or popular over-the-air televisions stations”); Government Reply at 10 (“these same stations are particularly weak financially”); Government Brief at 35 (discussing “marginal financial position of these stations”). Indeed, they stress their experts' observations that the audience appeal of the stations' programs (whether measured in number of listeners or intensity of preference) will not be enough to seriously enhance a cable operator's subscriber revenue (via more subscribers or higher fees). See, e.g., Noll Deposition at 60–63; id. at 142 (drop of third noncommercial outlet or weak UHF station will not affect decisions to subscribe); id. at 166 (similar); Second Declaration of Tom Meek ¶ 34. Plaintiffs have offered evidence that the average cable subscriber was served by a cable system carrying local broadcast stations accounting for about 97% of television ratings in non-cable households. Besen Decl. at 41; Plaintiffs' Joint Statement of Undisputed Facts Established by the Record (“Plaintiffs' Facts”) ¶ 9. While defendants bristle at this statistic, and suggest that there are defects in the type of Nielson data that support it, Defendants' Fact Response ¶¶ 7–9, they offer no alternative figure. And their own contentions—as to the marginality of electing stations and as to the improbability that they ...
110
Blake E. Reid @chup.blakereid.org · 13/05/2025
Great! Now repeat the exercise for other lawyers
16. With greater frequency, courts are now regularly evaluating the
conduct of lawyers and pro se litigants who improperly use AI in submissions
to judges. Whether that conduct supports the imposition of various types of
sanctions requires a fact- and circumstance-specific analysis. See, e.g., United
States v. Hayes, ___ F.Supp.3d ___, 2025 WL 235531 at *10-15 (E.D. Cal.
Jan 17, 2025) (sanctioning criminal defense lawyer for using AI; when
questioned by the court, the lawyer’s response about the source of inaccurate
legal citations “was not accurate and was misleading”); Saxena v. Martinez-
Hernandez, 2025 WL 1194003 at *2 and n.5 (D. Nev. April 23, 2025)
(“Saxena’s use of AI generated cases – and his subsequent refusal to accept
responsibility for doing so – is just another example of Saxena’s abusive
litigation tactics, and further explains why the court issued case-terminating
sanctions”) (collecting cases); United States v. Cohen, 724 F.Supp.3d 251, 254,
259 (S.D.N.Y 2024) (declining to find bad faith where defense lawyer
voluntarily disclosed that she “had been ‘unable to verify’” false citations in
colleague’s brief and lawyer acknowledged that he “would have withdrawn the
[fake] citations immediately if given the opportunity”).
242
Blake E. Reid @chup.blakereid.org · 13/05/2025
Well, this is a hell of a metaphor for generative AI
A final note. Directly put, Plaintiff’s use of AI affirmatively misled me.
I read their brief, was persuaded (or at least intrigued) by the authorities that
they cited, and looked up the decisions to learn more about them – only to find
that they didn’t exist. That’s scary. It almost led to the scarier outcome (from
my perspective) of including those bogus materials in a judicial order. Strong
deterrence is needed to make sure that attorneys don’t succumb to this easy
shortcut.
151120382
Blake E. Reid @chup.blakereid.org · 10/05/2025
Here’s the official explanation for the pre-publication. (My reaction: okay, but why now?)
The Office is releasing this pre-publication version of Part 3 in response to congressional
inquiries and expressions of interest from stakeholders. A final version will be published in
the near future, without any substantive changes expected in the analysis or conclusions.
1562
Blake E. Reid @chup.blakereid.org · 09/05/2025
Here's the DDC on that point (though in service of answering a somewhat different question) www.courtlistener.com/opinion/2400...
While the plaintiff is correct that the Library is codified under Title 2 of the United States Code, which addresses specifically the Legislative Branch, the court in Eltra found this irrelevant, holding that "such code-grouping cannot determine whether a given function is executive or legislative." 579 F.2d at 301. Moreover, the court in Eltra also noted that "[t]he Librarian performs certain functions which may be regarded as legislative (i.e., Congressional Research Service) and other functions (such as the Copyright Office) which are executive or administrative," and "[b]ecause of its hybrid character, it could have *43 been grouped code-wise under either the legislative or executive department." Id. But, the court found "such code-grouping" not to be dispositive, concluding "that the Copyright Office is an executive office, operating under the direction of an Officer of the United States and as such is operating in conformity with the Appointments Clause." Id.
110
Blake E. Reid @chup.blakereid.org · 09/05/2025
Here's one court concluding that the President has the power to remove the Librarian at will (albeit in dicta, and it's just the US DC District Court, not an appellate decision) www.courtlistener.com/opinion/2400...
 Most importantly, the Librarian is appointed by the President with the advice and consent of the Senate. 2 U.S.C. § 136. In addition, the President, not Congress, has the power to remove the Librarian at will.
241
Blake E. Reid @chup.blakereid.org · 09/05/2025
E.g., here's (my summary of) a district court getting wrapped around the axle about the LOC's branch of government in trying to figure out whether the APA applied to its rulemaking activity (ultimately overruled by the DC Circuit, but it illustrates the problem).
Nope, says the court (my paraphrase): the Library of Congress is a chameleon that can escape judicial scrutiny as a non-hybrid legislative thingy that is inseparable from Congress for the purposes of the APA and simultaneously escape separation-of-powers scrutiny by acting as a hybrid legislative-executive thingy for the purposes of the Constitution, and oh yeah, it’s the executive part of the hybrid thingy that’s doing the rulemaking for Constitutional purposes even though it’s the legislative whole of the non-hybrid thingy that’s doing the rulemaking for APA purposes.
121
Blake E. Reid @chup.blakereid.org · 09/05/2025
I know: "of Congress." But, which branch the LOC sits in is a pretty complex question. Here's Aaron Perzanowski detailing the mess. (Full article: scholarlycommons.law.case.edu/faculty_publ...)
28. Some courts have concluded that the Office is part of the legislative branch. See, e.g.,
United States v. Brooks, 945 F. Supp. 830, 834 (E.D. Pa. 1996) (“[T]he Copyright Office is
part of the legislative branch.”); Harry Fox Agency, Inc. v. Mills Music, Inc., 720 F.2d 733,
736 (2d Cir. 1983) (“[T]he Library of Congress . . . is a part of the legislative branch
itself . . . .”); Barger v. Mumford, 265 F.2d 380, 382 (D.C. Cir. 1959) (“[T]he Library of
Congress has long been treated as being in or under the jurisdiction of the legislative
branch . . . .”). Other courts have deemed the Office executive in nature. See, e.g., Eltra Corp.
v. Ringer, 579 F.2d 294, 301 (4th Cir. 1978) (“[T]he Copyright Office is an executive
office . . . .”); Intercollegiate Broad. Sys., Inc. v. Copyright Royalty Bd., 684 F.3d 1332, 1342
(D.C. Cir. 2012) (discussing why the Library of Congress “is undoubtedly a ‘component of the
Executive Branch’ ”). Acknowledging that “it is not clear whether the Library of Congress is
part of the executive or legislative branch,” the Ninth Circuit explicitly avoided the question
in Fox Television Stations, Inc. v. Aereokiller, LLC, 851 F.3d 1002, 1013 n.4 (9th Cir. 2017) (noting
that if it were to decide what level of deference the Office is owed, the court “would be
required to rule on constitutional questions that could have outsized consequences relative to
this case—such as determining whether the Library of Congress is a legislative or executive
agency”).29. Senator Orin Hatch, for example, has maintained that “the Copyright Office is in the
legislative branch of the Government, [and] whenever [it] is tasked with an executive-type
function, [a] constitutional question arises.” See John Duffy, Peter Strauss & Michael Herz,
Copyright’s Constitutional Chameleon, CONCURRING OPINIONS (May 17, 2013),
https://concurringopinions.com/archives/2013/05/copyrights-constitutional-chameleon
.html#more-74811 [https://perma.cc/NU7E-UUCQ]; see also H.R. REP. NO. 105-796, at 77
(1998) (Conf. Rep.) (referring to the Office as “a hybrid entity that historically has performed
both legislative and executive or administrative functions” and as “a legislative branch
agency”).
30. A 1996 Office of Legal Counsel memo cautioned that entities such as the Library of
Congress, “exercise authority that seems incompatible or at least difficult to reconcile with the
Supreme Court’s anti-aggrandizement decisions.” Constitutional Separation of Powers
Between the President and Cong., 20 Op. O.L.C. 124, 172 (1996).
3224