Education Copyright Checklist: Faculty Ownership, Classroom Use, Course Packs, Platform Terms, and Student Work

By Casey Scott McKay · 2026-07-13

This checklist audits and then builds copyright compliance at an educational institution across the four systems that actually matter. Phase one interrogates the ownership policy by category and tests whether it is an instrument or a webpage, because the teacher exception is custom rather than statute. Phases two and three build the four-question fair use workflow and apply it to reserves, course packs, film, images, and recorded lectures. Phase four reads the platform terms that quietly reallocate rights in material the institution does not own, and phase five covers student work, FERPA, and the accessibility programme with two independent legal bases. It closes on section 512 hygiene, training, escalation, and the four numbers worth reporting.

IP and Technology > Copyright | Checklist | Published 13 July 2026 - Updated 5 August 2026 | Casey Scott McKay - marksy.us

Summary. This checklist audits and then builds copyright compliance at an educational institution across the four systems that actually matter. Phase one interrogates the ownership policy by category and tests whether it is an instrument or a webpage, because the teacher exception is custom rather than statute. Phases two and three build the four-question fair use workflow and apply it to reserves, course packs, film, images, and recorded lectures. Phase four reads the platform terms that quietly reallocate rights in material the institution does not own, and phase five covers student work, FERPA, and the accessibility programme with two independent legal bases. It closes on section 512 hygiene, training, escalation, and the four numbers worth reporting.

Keywords: education copyright checklist · ownership policy audit · teacher exception · substantial resources threshold · signed instrument · fair use workflow · licence availability test · e-reserves review · course pack permissions · classroom exemption · lecture recording clearance · TEACH Act prerequisites · section 512 designated agent · repeat infringer policy · platform licence grant · exit and portability · accessibility procurement · student work licence · FERPA parallel analysis · programme metrics

How to use this checklist

| Field | Detail | |---|---| | Who runs it | Library or copyright office lead, with procurement, disability services, and counsel | | When | On standing up a programme; annually thereafter; and before any major platform adoption | | Time required | Three to six months for the first full pass | | Gates | Ownership policy executed; workflow embedded; vendor grants narrowed; agent registered | | Output | An executed policy, a running workflow with records, a reviewed vendor inventory | | Companion documents | Running Copyright Compliance at an Educational Institution and Teaching From Someone Else's Book |

The matter. A regional university with eleven thousand students is launching a fully online degree. The flagship course was designed and recorded by a professor now in negotiations with a competitor. The learning management system contract, signed four years ago by IT on price, grants the vendor a perpetual sublicensable licence in uploaded content. The library runs an e-reserves service on a remembered ten percent rule. Disability services buys converted textbooks from a vendor at considerable cost while the institution almost certainly qualifies as an authorised entity. Nobody has checked whether the designated agent registration is current. The provost wants to know if the online degree can launch.

Phase 1. Audit the ownership policy

[ ] Find the policy and date it. Why. A policy drafted before online course products and generative tools is silent on the categories that now generate the disputes. Trap. Assuming the intranet version is the operative one. Find the version that was approved and the version that anyone signed.
[ ] Test whether it is an instrument. Is it referenced in the employment agreement or appointment letter, and is that signed? Why. 17 U.S.C. § 204 requires a signed writing for a transfer of copyright ownership. Trap. A handbook page reciting institutional ownership. It transfers nothing.
[ ] Check that it addresses six categories separately. Traditional scholarship; teaching materials; recorded lectures and online course content; software and data sets; commissioned works; student work. Why. These have different answers and a policy addressing one addresses none of them. Trap. A single sentence about works created in the course of employment.
[ ] Read the substantial resources threshold. Why. If it captures an office, a laptop, a salary, and a sabbatical, it swallows the scholarship disclaimer entirely. Trap. Leaving it undefined and discovering at the dispute that both readings are available.
[ ] Confirm the policy states the institution's position on the teacher exception. Why. 17 U.S.C. § 101 and 17 U.S.C. § 201(b) point at institutional ownership on their face, Hays v. Sony Corp. of America questioned the exception and declined to decide it, and Weinstein v. University of Illinois routes the same result through the scope of employment. Custom is not a policy. Trap. Relying on what everyone assumes.
[ ] Check for reciprocal licences. Institution to faculty for continued teaching and scholarship; faculty to institution to keep running the course. Why. This is the settlement that prevents the disputes that actually happen. Trap. A one-way grant that guarantees the fight on departure.
[ ] Confirm the departing faculty position is stated expressly. May they teach the course elsewhere; may the institution run it without them. Why. Every real dispute in this area has that shape. Trap. Leaving it to be worked out at the exit interview.
[ ] Verify contractor assignments exist. Instructional designers, videographers, freelance authors. Why. Community for Creative Non-Violence v. Reid establishes that a contractor is not an employee, and the specially commissioned route needs both a written agreement and an enumerated category that instructional design does not obviously fit. Trap. Assuming a purchase order transferred copyright.
[ ] [Gate] Do not launch an online product until ownership of its components is established. Why. An institution that cannot show title cannot licence, cannot enforce, and cannot train on it.
[ ] Reconcile the copyright policy against the research and IP policy. Why. They overlap on software and data sets and are frequently inconsistent. Trap. Two policies, two owners, no reconciliation.

Phase 2. Build the fair use workflow

[ ] Frame it internally as a damages defence. Why. 17 U.S.C. § 504(c)(2) remits statutory damages entirely where a nonprofit educational institution employee acting within the scope of employment had reasonable grounds to believe the use was fair — and reasonable grounds is a documentation standard. Trap. Selling it as compliance overhead, which guarantees it is under-resourced.
[ ] Question one: is it already licensed to us? Why. A large share of what reaches the queue is covered by a database subscription permitting course use or persistent linking. Trap. Running a fair use analysis on material the library already licensed.
[ ] Question two: can it be licensed, and at what price? Why. Cambridge University Press v. Patton makes the availability of a licence for the specific excerpt directly relevant to the fourth factor. Trap. Skipping the check because permissions feel like a defeat.
[ ] Question three: how much, and behind what access control? Why. Cohort-limited, term-scoped, authenticated access does real work on the fourth factor. Trap. Posting to an open course page.
[ ] Question four: does the use serve the original's purpose or a different one? Why. Andy Warhol Foundation v. Goldsmith narrowed transformative use to that question, and assigning a chapter as reading serves the purpose the book was written for. Trap. Calling every pedagogical use transformative, which weakens rather than strengthens the position.
[ ] Retire the ten percent rule in writing. Why. It descends from the 1976 classroom guidelines, which were never enacted, were expressly framed as a minimum rather than a maximum, and were declined controlling weight in Cambridge University Press v. Patton. Trap. Replacing it with a different bright line. The Eleventh Circuit rejected the district court's substitute too.
[ ] Record each assessment. Work, portion, course, term, access control, the four answers, the decision, date, and person. Why. The record is the reasonable grounds. Trap. A decision with no written basis, which is legally worth nothing.
[ ] Embed the workflow where the decision happens. The reserves request form, the LMS upload flow, the course pack process. Why. A policy page requiring someone to go looking is not used. Trap. Building a beautiful guidance site nobody visits.
[ ] [Gate] Set escalation triggers. Substantial portion of one work; permission sought and refused; public rather than cohort-limited use; intent to license out; circumvention involved; purpose-built educational product; rightsholder already in contact. Why. Everything else must run without legal review or the system gets bypassed.

Phase 3. Course materials in operation

[ ] License course packs by default. Why. Princeton University Press v. Michigan Document Services and Basic Books v. Kinko's Graphics both turned on the established permissions market. Trap. Reading them as inapplicable because the defendants were copy shops. The market reasoning transfers.
[ ] Publish a course pack deadline early enough for permissions to clear. Why. The largest single driver of unlicensed copying is a reading list finalised two weeks before term. Trap. A deadline nobody enforces.
[ ] Run reserves per excerpt, not per policy. Limit to the enrolled cohort, scope to the term, take down at the end. Why. Cambridge University Press v. Patton requires a work-by-work analysis. Trap. A standing reserves collection that persists across years.
[ ] Prefer a durable link over a copy. Why. A link into a licensed database is not a reproduction at all. Trap. Scanning material the institution already subscribes to.
[ ] Set an internal escalation cap per work. Why. Not as a legal rule, but because heavy drawing from one book usually means the licensing analysis points the other way and the licence may be inexpensive. Trap. Publishing the cap as though it were law, which recreates the folklore.
[ ] Tell faculty they may show the whole film in a physical classroom. Why. 17 U.S.C. § 110(1) permits performance or display of a work in face-to-face teaching at a nonprofit institution from a lawfully made copy. No clip length, no licence. Trap. A copyright office that never says yes, which teaches faculty not to ask.
[ ] Explain what section 110(1) does not cover. Reproduction, distribution, and anything that leaves the room. Why. The exemption covers performance and display only, and it is tied to a place devoted to instruction. Trap. Ripping the film to a file and posting it under the classroom exemption.
[ ] Route online media three ways. Licensed streaming, a link to a service students already have, or limited clips with pedagogical framing under fair use. Why. Those are the routes that work. Trap. Defaulting to the fair use route because it requires no budget.
[ ] Run the circumvention analysis separately. Why. 17 U.S.C. § 1201 applies to defeating a technical measure independent of whether the underlying use is fair, and the triennial rulemaking exemptions carry conditions and expire. Trap. Assuming the educational exemption covers the institution's facts. Check the current text against the circumvention framework.
[ ] Handle recorded lectures as new works. Why. Recording creates a fixed work and leaves the face-to-face exemption behind, so every third-party work performed or displayed in the session needs its own analysis. Trap. Assessing each instance. Mature programmes edit third-party content out of recordings as routine, which is cheaper and more reliable.
[ ] Take guest speaker releases at the point of recording. Why. Retrospective clearance for a five-year-old recording costs the recording. Trap. A verbal agreement nobody wrote down.
[ ] Fund open educational resources properly if you want them. Release time and promotion credit. Why. Authoring an open textbook is a substantial uncompensated undertaking. Trap. Encouragement plus a dashboard, which produces a webpage.

Phase 4. Vendors and platforms

[ ] Inventory every platform holding institutional content. LMS, courseware, lecture capture, assessment, proctoring, plagiarism detection, and the departmental tools nobody registered. Why. You cannot review what you have not listed. Trap. Counting only what procurement processed.
[ ] Read the licence grant in uploaded content. Why. Perpetual, irrevocable, sublicensable, or improvement-of-the-service language covers faculty course materials and student submissions alike. Trap. Accepting a grant framed as a general right to use content rather than as what is necessary to operate the service.
[ ] Check whether the grant survives termination. Why. Many do, which means the vendor retains rights in years of faculty material after the institution leaves. Trap. Negotiating price and term and never reading the survival clause.
[ ] Test the exit. What comes out, in what format, and can anyone open it? Why. A platform whose export is an unopenable proprietary bundle has acquired the institution's teaching materials without buying them. Trap. Treating portability as an IT concern rather than an IP one. The technology contract analysis applies here in full.
[ ] Limit the institution's warranty of rights. Why. If faculty own their teaching materials and students own their submissions, the institution cannot validly grant the vendor a licence in either. Trap. A broad warranty that the institution has all necessary rights, which is a representation it cannot support.
[ ] Require AI training use to be opt-in and a material amendment. Why. Such provisions are now common, frequently live in an updatable policy rather than the signed agreement, and reach material the institution has no right to contribute. Trap. A change-of-terms clause that lets the vendor add it by notice.
[ ] Make accessibility a contract term with a named standard and remediation timelines. Why. A platform that cannot deliver accessible content forces conversion work a contract clause would have avoided. Trap. Accepting a roadmap commitment.
[ ] Run the privacy layer in parallel. FERPA, the applicable state privacy statutes, and for institutions serving minors the children's privacy regime. Why. A copyright-only review misses most of the exposure in an edtech contract. Trap. Two reviews that never meet.
[ ] [Gate] Route every adoption through one desk. Why. Departmental adoption of free tools is how an institution acquires forty unreviewed vendor relationships.

Phase 5. Student work, accessibility, and the library

[ ] State in policy that students own their work. Why. A student is not an employee; authorship vests on fixation under 17 U.S.C. § 102. Trap. A professor who assumes work produced in their course is theirs.
[ ] Take a narrow licence at enrolment. Non-exclusive; reproduction and display for educational, assessment, accreditation, and internal institutional purposes. Why. It covers the routine uses. Trap. A broad grant that reads as an overreach and invites the complaint.
[ ] Add two separate opt-ins. Public showcase and marketing; anything commercial or model training. Why. Bundling them taints the whole licence. Trap. Burying them in an enrolment click-through nobody reads, which is also the assent design problem.
[ ] Check the plagiarism service's retention terms. Why. Retaining submissions in a comparison database is a reproduction and its use is a distribution question. Trap. Relying on someone else's litigated fair use outcome instead of obtaining consent.
[ ] Read the deposit agreement actually in force for theses. Why. They vary enormously by era and are frequently narrower than an open access mandate assumes. Trap. Applying today's mandate to a 1998 deposit.
[ ] Run FERPA as a parallel analysis. Why. A copyright licence to display work does not authorise disclosure of an education record, and where the two conflict the privacy answer governs. Trap. Treating consent to display as consent to disclose.
[ ] Confirm authorised entity status under 17 U.S.C. § 121. Why. It permits reproduction and distribution of previously published literary works in accessible formats for eligible persons, with no permission, no licence, and no portion limit. Most institutions with a disability services office qualify. Trap. Buying converted texts from a vendor at cost while holding the exception.
[ ] Note the second basis. Authors Guild v. HathiTrust held the same activity fair use independently. Why. Two independent bases is as good as a copyright position gets. Trap. Treating accessibility as the riskiest programme when it is the safest.
[ ] Consider 17 U.S.C. § 121A for international programmes. Why. Marrakesh implementation permits cross-border exchange with authorised entities abroad.
[ ] Produce genuinely accessible formats. Structured text, described images, tagged documents. Why. Image-only scans satisfy a checkbox and no reader. Trap. Measuring conversion volume rather than usability.
[ ] Use 17 U.S.C. § 108 for library preservation and replacement, and fair use for access. Why. Section 108 expressly preserves fair use, and its digital copies may not go beyond the premises. Trap. Citing section 108 as authority for a public collection.

Phase 6. Section 512, the TEACH Act prerequisites, and governance

[ ] Register a designated agent with the Copyright Office and diarise the renewal. Why. 17 U.S.C. § 512 safe harbour requires it, and registrations lapse on a schedule. Trap. A registration naming a person who left in 2019.
[ ] Adopt and actually implement a repeat infringer policy. Why. The safe harbour requires reasonable implementation, not adoption, and the cases turn on application. Trap. A policy that has never been applied to anyone.
[ ] Run notices and counter-notices on the statutory timetable. Why. The procedural mechanics are unforgiving. Trap. Informal handling that satisfies nobody and preserves nothing.
[ ] Use section 512(e) and know its limits. Why. It limits liability for faculty and graduate students acting as instructors or researchers, conditioned on providing informational materials describing and promoting copyright compliance. Trap. Missing the exclusions — instructional materials required or recommended in the preceding three years, and more than two notices about the same person — which bite exactly where the protection is most wanted.
[ ] Meet the TEACH Act institutional prerequisites even though you will not rely on it. Instituted policies, informational materials, notice to students. Why. They overlap with the section 512(e) conditions and with System One. Trap. Building a programme on 17 U.S.C. § 110(2), whose technological measure requirements, portion limits, and exclusion of purpose-built digital instructional products make it narrower than fair use.
[ ] Register the institution's own works. Owned course products, commissioned content, software, publications. Why. 17 U.S.C. § 411 makes a completed registration a precondition to suit, and timely registration preserves statutory damages and fees. Trap. Looking for the registration after discovering the copying.
[ ] Centralise rightsholder contact. One address, one owner, restrict access first and verify the claim second. Why. Most contacts resolve with a removal, a licence, or a correction; an ignored one escalates. Trap. Letting each department field its own.
[ ] Train to the four questions, not the doctrine. Why. The decision-makers are instructors, librarians, and designers. Trap. A four-factor lecture that leaves everyone less confident than before.
[ ] Publish response times and hold to them. Why. A six-week permissions turnaround in a twelve-week term is a refusal expressed differently. Trap. Measuring accuracy and not speed.
[ ] [Gate] Report four numbers annually. Permissions processed and turnaround; assessments completed; vendor grants reviewed and narrowed; rightsholder contacts and resolutions. Why. They show whether the system is used or routed around. Trap. Reading a near-empty queue as low risk. At a large institution it means the teaching is happening off the books.

Phase 7. Working the sample matter

[ ] Stop the online degree launch pending a title review of the flagship course. Why. The professor designed and recorded it and is in negotiations with a competitor; if the policy is a webpage rather than an instrument, the institution's position on the recording is far weaker than the provost assumes. Trap. Launching and then discovering the title question in a cease and desist.
[ ] Separate the flagship course into its components. Underlying scholarship and lecture content; slides and problem sets; the recording itself; the instructional design; the platform packaging. Why. Five components, potentially four owners. The recording may be a derivative work of content the professor owns. Trap. Asking "who owns the course," which has no answer.
[ ] Find the videographer's and instructional designer's contracts. Why. Community for Creative Non-Violence v. Reid means contractors own absent a written assignment, and video production and instructional design do not sit comfortably in the specially commissioned categories. Trap. Assuming a purchase order did the work.
[ ] Negotiate the reciprocal licence now, while there is still a relationship. Institution keeps running the recorded course; professor keeps teaching their own scholarship elsewhere. Why. This is the settlement both sides actually need, and it is far cheaper before the departure than after. Trap. Standing on a policy position nobody has tested and forcing the fight.
[ ] Read the four-year-old LMS contract's licence grant. Why. A perpetual sublicensable grant in uploaded content covers the flagship recording, every other faculty member's materials, and every student submission — none of which the institution clearly owns. Trap. Assuming a signed contract is a settled contract. Ask for an amendment at renewal, and price the switch as leverage.
[ ] Test the LMS export before the renewal conversation. Why. If the export is unopenable, the institution has no alternative and no leverage. Trap. Discovering that during a migration.
[ ] Audit the e-reserves service against the four questions. Why. A remembered ten percent rule produces no records, which means no reasonable grounds under 17 U.S.C. § 504(c)(2) and no defence. Trap. Cutting the service instead of documenting it, which sends the copying underground.
[ ] Check whether disability services needs the vendor at all. Why. If the university is an authorised entity under 17 U.S.C. § 121, it may make accessible copies itself with no permission and no licence, backed independently by Authors Guild v. HathiTrust. Trap. Continuing to buy what the institution is entitled to produce, which is a budget problem disguised as a compliance one.
[ ] Look up the designated agent registration today. Why. It takes five minutes and an expired registration means no 17 U.S.C. § 512 safe harbour at all. Trap. Assuming IT keeps it current.
[ ] [Gate] Give the provost a staged answer. The degree can launch on components with clear title; the flagship course launches once the reciprocal licence is executed; the LMS grant is renegotiated at renewal. Why. A blanket no gets overruled and a blanket yes creates the claim. Trap. Answering the question asked rather than the question that matters.

Phase 8. Generative AI, in three separate questions

[ ] Question one: what does the institution feed to vendors? Why. Training-use provisions are now standard, frequently sit in an updatable policy rather than the signed agreement, and reach faculty and student material. Trap. Debating AI policy in the abstract while the LMS terms already grant it.
[ ] Question two: what do faculty and students produce with these tools? Why. The Copyright Office's human authorship position excludes purely machine-generated material from protection, which matters for institution-owned course products and for thesis deposits. Trap. Merging this with academic integrity, which is a different policy owned by different people.
[ ] Question three: what does the institution build on its own corpus? Why. A tutoring or research tool trained on course materials, recorded lectures, and student submissions is lawful only if the institution holds the rights, and Phase 1 and Phase 5 determine whether it does. Trap. Building first. An institution that never resolved ownership of recorded lectures cannot train on them, and retrospective consent across graduated cohorts is not obtainable.
[ ] Read the vendor indemnity now rather than later. Why. Where a rightsholder asserts that the institution's use of a tool infringed, the analysis turns on what the institution did, and the indemnity is the only thing that shifts it. Trap. Discovering the carve-outs during the claim.
[ ] Assume any policy drafted before 2020 is silent on all three. Why. It is. Trap. Waiting for a proposal that forces the issue instead of closing the gap at the next revision.

Phase 9. Adapting this for K-12

[ ] Recognise that the ownership analysis flips. Why. A schoolteacher delivering a set curriculum with district-specified materials looks far more like a conventional work made for hire than a professor writing a monograph, and the academic freedom framing behind the teacher exception is largely absent. Trap. Copying a university policy that disclaims ownership in scholarship into a district that has no scholarship.
[ ] Decide deliberately about lesson plan marketplaces. Why. Teachers now sell materials created on district time using district templates, and most districts have no position on it. Trap. Taking a position for the first time in a dispute with a teacher who has been selling for six years.
[ ] Weight the privacy layer heavily. FERPA, the children's privacy regime, and state student data statutes. Why. In K-12 this is usually larger than the copyright exposure. Trap. Running a copyright-only vendor review.
[ ] Make vendor review the first system, not the third. Why. Districts adopt platforms centrally, on price, for multi-year terms, across every school. The blast radius of one unreviewed contract is the whole district. Trap. Sequencing as though the institution were a university.
[ ] Confirm the damages remission applies. Why. Public school districts are nonprofit educational institutions for 17 U.S.C. § 504(c)(2) purposes, on the same documented good-faith condition. Trap. Assuming the protection without building the documentation that triggers it.
[ ] Replace the folklore in the staffroom too. Why. The ten percent rule is repeated there with the same confidence and the same absent source. Trap. Assuming a training session fixed it. Put the four questions in the request form.

Phase 10. Sequencing, resourcing, and the risk conversation

[ ] Sequence twelve months in dependency order. Months one to three, the ownership policy through governance and execution. Months two to four, the fair use workflow embedded in the reserves and upload flows. Months four to six, vendor inventory and licence grant review. Months six to nine, accessibility and library exceptions. Months nine to twelve, section 512 hygiene, student licensing at enrolment, training, and the first annual report. Why. Each system makes the next cheaper, and the policy is the longest lead item because it needs consultation. Trap. Starting with training, which is the visible one and the one that depends on everything else.
[ ] Resource it as roughly one full-time equivalent in a mid-sized institution, located in the library, with a named counsel contact for escalation. Why. Institutions that distribute it as a fraction of five roles produce a policy document and no system. Trap. Assigning it to whoever asked about it.
[ ] If the budget supports only three things, choose these. The ownership policy as an executed instrument; the four-question workflow embedded in the reserves process; the vendor licence grant review. Why. Those three cover the disputes that actually occur. Trap. Spending the budget on a guidance website.
[ ] Answer the exposure question with structure rather than reassurance. Statutory damages are largely remitted for documented good-faith uses under 17 U.S.C. § 504(c)(2); state institutions retain substantial sovereign immunity protection against damages in federal court while private institutions have no equivalent; injunctive relief is available against everyone. Why. The realistic remedy is an injunction, a licence negotiated under pressure, and costs. Trap. Catastrophising, which produces overreaction and underinvestment at once.
[ ] Name the larger exposures explicitly. A vendor grant of rights the institution did not hold. A platform exit that stranded years of course content. An accessibility failure that produced a complaint. A departing faculty dispute over a flagship programme. Why. None are infringement claims and all are copyright problems, which is the argument for building all the systems rather than only the one that answers the fear. Trap. Letting the risk conversation stay about hypothetical damages.
[ ] Diarise the annual pass. Re-read the policy against new categories; re-check the agent registration; re-inventory the platforms; re-run the four numbers. Why. Every input changes on a schedule. Trap. Filing this as a completed project.

Phase 11. Protecting what the institution owns

[ ] Register owned course products, commissioned content, software, and publications. Why. 17 U.S.C. § 411 makes a completed registration a precondition to suit for United States works, and timely registration preserves statutory damages and fees. Trap. Going looking for a registration after a competitor has copied the online programme wholesale.
[ ] Keep the executed instrument in the file for every owned work. Why. Ownership asserted in policy and unsupported in the file is not ownership you can prove. Trap. A chain of title that exists only in institutional memory. Run the chain of title audit once and it will find the gaps.
[ ] Close the contractor gap systematically. Freelance instructional designers, videographers, editors, illustrators, translators. Why. Each owns their contribution absent a written assignment, and the specially commissioned route needs both a signed agreement and a fitting statutory category. Trap. Fixing this one contract at a time as disputes surface.
[ ] Ask who runs the trademark portfolio. Institution name, seal, athletic marks, programme names. Why. They are trademarks, frequently licensed, frequently unregistered, and they are not the copyright office's job — but somebody should confirm they are somebody's. Trap. Assuming the licensing office is registering what it licenses.
[ ] Reconcile with the research IP framework. Why. Sponsored research, material transfers, and inventions sit under grant terms and the technology transfer regime, not under the copyright policy, and the two overlap on software and data sets. Trap. Two documents giving different answers on the same data set, discovered when a spin-out needs a clean licence.
[ ] [Gate] Confirm the institution can grant what it promises before signing any licence out. Why. An institution licensing its online curriculum to a partner warrants title it may not hold. Trap. Warranting first and auditing after.

[ ] Make the copyright office useful before making it authoritative. Why. The most common institutional failure is a service that says no by default; faculty respond by not asking, which removes the institution's visibility and its documentation simultaneously. A service that finds the licence, finds the open alternative, or documents the fair use gets used. Trap. Measuring the office by refusals rather than by turnaround.
[ ] Write down what the programme deliberately is not doing. Item-level clearance of legacy reserves, retroactive clearance of old recordings, and pursuit of every unreviewed departmental tool at once. Why. A documented accepted risk is a governance decision; an undocumented one reads later as an oversight. Trap. Leaving the omissions implicit.

[ ] Name a successor for every standing obligation. The rightsholder contact address, the agent registration renewal, the annual metrics, and the vendor renewal diary. Why. Each outlives the person who set it up, and an obligation with no named successor lapses silently. Trap. A handover document listing systems rather than promises.
[ ] Re-execute the policy on material amendment. Why. A signed instrument covers the version signed; adding a new ownership category by administrative update does not reach anyone who signed the old one. Trap. Amending the webpage and assuming the signatures followed.

[ ] Give instructors the three-line version they can act on. Check whether the library already licenses it; ask the copyright office before the term starts; and know that in a physical classroom you may show the whole film. Why. Most faculty will never read the policy and all of them will remember three lines. Trap. Distributing the full framework and calling that training.

[ ] Then put those three lines on the reserves form itself, where the decision is being made. Why. Guidance at the point of action is the only guidance that gets applied.

Outcome. An executed ownership policy addressing six categories with reciprocal licences and a defined resources threshold. A four-question workflow embedded where decisions are made, generating the records that support remission under 17 U.S.C. § 504(c)(2). A vendor inventory with narrowed grants, tested exits, and contractual accessibility. A student work licence with separate opt-ins and a parallel privacy analysis. An accessibility programme resting on a statutory exception and a Second Circuit holding. A current designated agent, an implemented repeat infringer policy, and four numbers the provost can read.

The five things people get wrong

Treating the policy as a webpage. 17 U.S.C. § 204 requires a signed writing. An intranet page asserting institutional ownership transfers nothing, and the departing-professor dispute is where that is discovered.

Repeating the ten percent rule. It is not in the statute, was never enacted, was framed as a minimum, and was declined controlling weight in Cambridge University Press v. Patton. It suppresses lawful teaching while protecting nobody.

Relying on the TEACH Act. 17 U.S.C. § 110(2) demands technological controls most systems do not implement, limits portions for exactly the media people want, and excludes purpose-built digital instructional products. Fair use is broader and lighter.

Signing platform terms that grant rights the institution does not hold. If faculty own their materials and students own their submissions, a warranty of all necessary rights is unsupportable — and the vendor's perpetual sublicensable grant reaches both.

Funding accessibility last. It has a statutory exception under 17 U.S.C. § 121 and an independent fair use holding in Authors Guild v. HathiTrust, the least exposure of anything in the programme, and the readers with the greatest need.

Key Authorities at a Glance

| Authority | Proposition | |---|---| | 17 U.S.C. § 101 | Work made for hire definition | | 17 U.S.C. § 102 | Subject matter; authorship on fixation | | 17 U.S.C. § 106 | Exclusive rights | | 17 U.S.C. § 107 | Fair use | | 17 U.S.C. § 108 | Libraries and archives | | 17 U.S.C. § 109 | First sale | | 17 U.S.C. § 110 | Face-to-face teaching; TEACH Act | | 17 U.S.C. § 121 | Accessible formats | | 17 U.S.C. § 121A | Cross-border accessible copies | | 17 U.S.C. § 201 | Ownership; work made for hire | | 17 U.S.C. § 204 | Signed writing for transfers | | 17 U.S.C. § 411 | Registration precondition to suit | | 17 U.S.C. § 504 | Damages; nonprofit remission | | 17 U.S.C. § 512 | Safe harbour; designated agent; section 512(e) | | 17 U.S.C. § 1201 | Circumvention; triennial exemptions | | Cambridge University Press v. Patton | E-reserves; work-by-work analysis | | Princeton University Press v. Michigan Document Services | Commercial course packs | | Basic Books v. Kinko's Graphics | Copy shop course packs | | Andy Warhol Foundation v. Goldsmith | Transformative purpose narrowed | | Campbell v. Acuff-Rose Music | Transformative use framework | | Authors Guild v. HathiTrust | Search and accessibility as fair use | | Authors Guild v. Google | Snippet display for search | | Community for Creative Non-Violence v. Reid | Contractor is not an employee | | Hays v. Sony Corp. of America | Teacher exception questioned | | Weinstein v. University of Illinois | Academic authorship custom | | FERPA | Education records privacy | | Section 1201 educational exemptions | Triennial rulemaking | | Classroom guidelines legislative history | The unenacted guidelines |