Copyright Act Panel 4: The Shifting Line Between Federal and State Protection

29th Annual BTLJ-BCLT Spring Symposium: Origins, Evolution, and Possible Futures of the 1976 

R. Anthony Reese, UCI School of Law

Fundamental change: eliminating common-law copyright for unpublished works and unifying the regime at creation. Contemporaries like Ralph Sharp Brown saw it as a huge, pivotal change. Now we take it as easy background principle.

Federal law did provide a cause of action before 1909 for unpublished copying of unpublished manuscripts. But it was a procedural door, not substantive. More significantly, 1909 Act dropped that but did allow certain types of unpublished works to obtain federal © by registration. Categories where works were commonly performed/exhibited rather than being published. Most people think of this as a footnote, but this new option turned out to mark a significant shift in the state/federal protection divide: for lecture, dramatic/musical composition, motion picture, photograph, work of art, drawing.

Registration data 1929-77: 25% of total nonrenewal registrations were for unpublished works. That’s a big deal! Understates importance because of the limit on classes of eligible works. 28% of all registrations were musical works, and 83% of those were unpublished works by 1977 though it took many years to get there. Similar story with drama registrations (88% of those were unpublished). For other classes it was less significant (except for lectures, at 100%); 46% of scientific drawings and 29% of photographs from 1954-1977 that were registered were unpublished.

Office turned down lots of requests to register unpublished books. There was also uncertainty about what constituted publication. And perpetual state law protection through broadcast, performance, and even potentially distributing phonorecords was a worry: owners could economically exploit their works in front of millions w/o a © bargain/any duration endpoint.

Considered alternatives: extend voluntary registration to all classes; make public dissemination not publication the dividing line (“communicating a work to the public visually by any acoustically by any method and in any form); or eliminate state protection for unpublished works and provide federal © from creation. There were various views. Learned Hand favored the middle alternative for undisseminated works, provided there was some time limit on state law protection. Concern w/infinite duration was supplemented w/concern that there was no fair use under state law (which no case ever held but there was speculation), and concerns about evading the compulsory mechanical license.

When they chose the final option, they did not apply any national origin rules; 105 was adjusted to extend to both published and unpublished works of the US gov’t. And of course duration rules had to change; added to the push for life+50 once dissemination could no longer serve as the universal starting point for a fixed term. Had to figure out what to do with pre-1972 sound recordings too; didn’t get taken up into federal © then b/c of larger lack of certainty about the topic.

Unchanged substantive law: © attached automatically on creation, w/o formalities, but publication would enter public domain unless formalities were complied with. Transfers also changed: divisible, writing required, recordation, subject to termination. Improved nonmonetary remedies. Every unpublished work by every person who died long enough ago now enters the public domain—the initial group in 2002 was the largest ever expansion of the public domain. All the statutory limits now apply to unpublished works. Statutory damages and attorneys’ fees for post-infringement registration too. The rights are 106 rights despite suggestions in common law that “any use” would infringe; idea/expression applies despite suggestion in English law that describing the Queen’s unpublished drawings would infringe; transferring the only copy would no longer transfer the ©.

There are still a lot of registrations of unpublished works: 38% of all registrations from 1978-2022. Mostly monographs, 27%; 65% of performing arts, 38% for visual arts; 64% for sound recordings. These registrations are no longer necessary but provide the advantages of registration.

Subsequent developments: clarified that fair use covered unpublished works; resolved split about whether sale of pre-1972 phonographs was publication, for musical works first and then literary/dramatic works. Finally brought pre-1972 sound recordings into sui generis protection, closing the circle/finally removing the last bit of rubble from the 76 Act’s destruction of the wall between published/unpublished works—nothing left for common law copyright to protect in fixed works.

Marketa Trimble, UNLV William S. Boyd School of Law: Would we expect state law diversity? Legislative laboratory?

Some preempted, often after a long time: Cal resale royalties enacted 76, held preempted 2018. NY standardized testing act; PA Feature Motion Picture Fair Business Practices Law. State statutes protecting rights to unfixed performances ok. Also, gov’t edicts doctrine matters to state law—state can’t claim © of materials produced in course of duties of judges and legislatures.

State laws are very outdated. Typically often list “copyrighting,” “causing to be copyrighted,” “acquiring copyright,” or “securing copyright” as distinct acts, rather than “registering.” Plenty of state laws assert “copyright” to state laws, and other types of works such as works developed by a county board of education (CA), data processing software created by gov’t agencies; and geological/topological surveys of PA. Also an Arkansas history textbook; official insignia for MD farm products; highway maps of Ohio; and the Oregon Blue Book—insight into what legislators feel a need to claim.

Since Ga v. Public.Resource.Org, some states have eliminated their provisions—NY, MD, ME, and Montana and OK already didn’t. VA authorizes releasee of all potentially ©able materials under a CC or Open Source Initiative license.

17 USC 1401 specifies that the person who has the exclusive right to reproduce a sound recording under the laws of any State as of the date before the date of enactment is the federal owner of pre-1972 sound recording rights. Recognizes that state laws varied on ownership. But what if the state laws conflict? There is no choice of law provision, and state laws do vary. Mostly it’s the label, whether framed that way in state decisions (CA) or statutes (AZ). WV has a different rule: label unless there’s no written contract, in which case it’s performers.

State law can also enhance protection of authors: Cal law purports to require generative AI developers to post a summary of the datasets used to develop a system, including sources/owners. MD bill prohibits inclusion in contract for state public art a requirement that artist waive moral rights under VARA. Another bill limits admissibility in criminal/juvenile proceedings of uses of “creative expression of a defendant or respondent.” Court is supposed to figure out whether expression was literal or fictional. Not admissible for mens rea, but could be used to decide on referral to mental health services/diversion programs.

Could protect users too, for example by preventing © as means to prevent access to public records: a series of new state bills require, e.g., public access to “learning materials.” Digital lending of e-books; MD’s bill was invalidated as preempted, but CT adopted a new bill prohibiting libraries from entering into contracts or license agreements for ebooks and digital audiobooks that contain certain restrictions.

Some states have considered laws on what demand letters should look like—abuse of rights provisions. And NV has an act creating a requirement for law enforcement agencies to adopt written policies and procedures governing performance of © works by peace officers while on duty. (B/c of a bizarre technique used where cops blasted music in the false belief that this would cause videos of their behavior to be unpostable online.)

Generative AI legislation: owners of the model training/generated content by the person who provides input; NY’s AI transparency for journalism act: disclosure obligation for content accessed by AI developer crawlers, for identity of crawlers.

Last example: help enforcing copyright outside of the US. Wash, LA, Utah—not clear how they fare after recent SCt decisions.

Are state legislators becoming more ambitious? Public becoming more attuned to ©

Guy Rub, Temple University School of Law: Contracts—“breach of contract” was removed from the text of 301, which now needs interpretation. Leg. History says that “equivalent to ©” was intended to be “the clearest and most unequivocal language possible, so as to foreclose any conceivable misinterpretation.” Whoops.

What is so hard about “equivalent”? © is entangled w/state commercial law. Contracts can ignore subject matter, rights, and defenses. Also depends on view of purpose of ©: delicate balance of competing interests of society and authors, v. exclusive right for benefit of authors.

Most common litigated pattern is idea submission; others are B2B transactions. After that there’s a lot of variety—contracting around fair use is extremely rare and mostly limited to reverse engineering. Is a promise an extra element? It’s a formalistic test. Courts also say that not every technical difference will suffice—must create meaningful distinction. Has to be about the nature of the cause of action.

Two approaches: majority: contracts are bilateral and thus not equivalent to property rights; versus minority approach: no, contracts can’t regulate actions that are exclusive rights under ©. ProCD is the most famous majority approach case, but not the first. Most appellate courts adopt it one way or another; only the 6^th Circuit explicitly rejected until recently.

Why so popular? It’s easy to apply and there are no great alternatives. But then the Second Circuit decided the Genius/Google case where Genius had browsewrap saying you can’t scrape; Google won. The contract limits reproduction/public display and is thus equivalent to © and preempted. When Genius sought cert, the SG argued that browsewrap was different (implicitly, not real consent).

Why does it matter? B/c of plenty of other attempts to limit scraping with browsewrap.

Conflict preemption might be the answer! Section 301(a) might ask the wrong question; interference w/the (c) system/obstacle to the goals of Congress. In re Jackson (2d Cir. 2020) (citing yours truly and Jennifer Rothman). Look at what the state is trying to promote: privacy or creativity/commercialization of information? Look at whether there’s harm to (c) law.

X Corp v. Bright Data (ND Cal 2024) found claims based on mining and sale of data are conflict preempted, b/c the interest is monetization, which is the same as ©; the harm is clear b/c it prevents users from commercializing their posts and b/c it circumvents fair use.

Conflict preemption is better than formalism: you can ask whether the contract was individually negotiated; you can ask about market power; you can ask about the purposes of the contract and of the alleged breacher.

Shyamkrishna Balganesh, Columbia Law School (Speaker and Moderator): Hot news had an outsized influence on 301. A misleading account continues to influence how courts talk about misappropriation to this day.

Misappropriation falls into disfavor after INS v. AP: effect of Brandeis’s dissent/Learned Hand’s refusal to expand or adopt the decision along with Erie v. Tompkin’s rejection of general federal common law. But states either through statute or common law began to absorb it. Legislative history suggests that misappropriation is structurally different from © and would allow equity to go down new paths. But the enacted version of the Act doesn’t contain the list containing misappropriation referred to in the legislative history.

Register’s 1965 report saw misappropriation as “the virtual equivalent of ©.” So proposed exclusions for contract, breach of trust, privacy, defamation, and deceptive trade practices, but not misappropriation. Then the Dep’t of Commerce intervened, in the form of a PTO rep, and said misappropriation was important b/c it allows courts to anticipate new areas for development and retain equitable flexibility. But Commerce hadn’t consulted w/other departments. “we have the Dep’t of State disagreeing w/everybody except on the manufacturing clause and now we have the Dep’t of Commerce that takes a different view. Does anyone purport to speak for the administration?” A staffer asks the DOJ to weigh in, and the DOJ is firmly opposed to misappropriation. It would neutralize the logic behind preemption. He says DOJ misrepresented misappropriation as creating antitrust concerns, simulating property rights, and too vague. [Honestly I don’t see that as a misrepresentation!]

Striking the provision’s list in full then eliminates the record of the logic.

There is a fundamental mismatch b/t the House Report and the actual law. Different courts of appeal seeking to resurrect INS in the 80s and 90s unfortunately rely on the old House Report saying that “based on legislative history, it is generally agreed that hot news survives preemption.” NBA v. Motorola. Only corrected in 2011 in Flyonthewall, but it moved to other jurisdictions like Ohio in the meantime, without being corrected there.

If I’d had time, I had a Q for Trimble: what are your thoughts about criminal anti-camcording and anti-copying rules for people who don’t own the masters, including laws making it illegal not to put the name of the actual copy maker on the copies? OCGA § 16-8-60 prohibits transferring “any sounds or visual images … onto any other … article without the consent of the person who owns the master” and separately makes it unlawful to sell any article on which sounds or visual images have been transferred “unless such … article bears the actual name and address of the transferor of the sounds or visual images in a prominent place on its outside face or package.”