1 of 44

Policy Implications for Colleges at the Crossroads of AI and IP Law

ASHLEY SLOAT, Ph.D. | MALLORY KING, J.D.

OCTOBER 14, 2025

This presentation is for information purposes only and does not constitute legal advice.

2 of 44

Ashley Sloat, Ph.D.�President & Director of Patent Strategy

3 of 44

Mallory King, J.D.�Owner & Attorney

4 of 44

Where are we now?

YOU ARE

HERE

AI-generated image with DALL·E 2

Rules-based AI follows predefined instructions and decision trees to handle inputs

Generative AI uses machine learning models to create new outputs by recognizing patterns and relationships in data

5 of 44

ChatGPT passed all three parts of the United States medical licensing examination �within a comfortable range

Source: Performance of ChatGPT on USMLE: Potential for AI-Assisted Medical Education Using Large Language Models

6 of 44

GPT-4 can write code in all major programming languages

7 of 44

GPT-4 passed the Uniform Bar Exam …and scored in the 90th percentile

Source: Katz, Daniel Martin and Bommarito, Michael James and Gao, Shang and Arredondo, Pablo, GPT-4 Passes the Bar Exam (March 15, 2023)

8 of 44

9 of 44

Source: Visual Capitalist | AI vs. Humans: Which Performs Certain Skills Better?

10 of 44

11 of 44

Overview

IP as constitutional rights
AI versus Human Creation of IP
Common Legal Pitfalls for Higher Ed with AI use
AI Impacts on Specific Educational Areas
Best Practices for Higher Ed

12 of 44

Overview

IP as constitutional rights
AI versus Human Creation of IP
Common Legal Pitfalls for Higher Ed with AI use
AI Impacts on Specific Educational Areas
Best Practices for Higher Ed

13 of 44

to promote the progress of science and useful arts, by securing, for a limited time, to authors and inventors, the exclusive Right to their respective writings and discoveries

“

– U.S. Constitution 1789

14 of 44

Intellectual Property Types

Acme

Widgets ™

Utility

Patent

Design

Patent

Copyright

Trademark

Trade Secret

15 of 44

Intellectual Property Types

Acme

Widgets ™

Utility

Patent

Design

Patent

Copyright

Trademark

Trade Secret

16 of 44

Patent Types

Plant

Design

Utility

Length: 15– 20 years

Enforcement: File patent app with USPTO; can enforce after issuance

17 of 44

Protects: Creative works of expression, i.e. books, music, art, photography, and software.

Length: Life of the author plus 70 years

Enforcement: Register with Copyright Office to be able to enforce it in court.

18 of 44

Trademarks

Protects: Names or logos used in association with goods/services.

Length: Indefinitely, if renewed properly.

Enforcement: Register with the USPTO for ® mark and access to federal courts.

19 of 44

Trade Secrets

Length: Indefinitely

Enforcement: Requires enterprise to take steps to keep the information secret

Example: Coca-Cola Recipe

[ARS]

The final non-patent related IP asset type is a trade secret.

A trade secret can refer to confidential, commercially valuable information that provides a business with a competitive edge. Unlike patents, trademarks, and copyrights, which all require public disclosure, trade secrets provide the competitive advantage of being kept confidential.

Trade secrets can protect things like formulas, recipes, algorithms, manufacturing processes, sales methods, client lists, business methods, and other proprietary information that gives a business an advantage over competitors. The formular for Coca Cola is perhaps the most famous trade secret.

A trade secret only gives protection if you keep it secret and only provides protection from:

Unauthorized disclosure and use of the trade secret by others

Obtainment of the trade secret by some improper means

They are generally protected under state law. Protection lasts indefinitely as long as it is kept secret. And if somebody independently derives it, you are out of luck

20 of 44

Brand Protection

The copyrights, trademarks, and contracts necessary to protect your brand’s rights and assets.

21 of 44

Overview

IP as constitutional rights
AI versus Human Creation of IP
Common Legal Pitfalls for Higher Ed with AI use
AI Impacts on Specific Educational Areas
Best Practices for Higher Ed

22 of 44

to promote the progress of science and useful arts, by securing, for a limited time, to authors and inventors, the exclusive Right to their respective writings and discoveries

“

– U.S. Constitution 1789

So can AI be an author or inventor?

23 of 44

Creation & Invention Limited to Humans

The term “individual” ordinarily means a human being, and Congress did not provide any indication it intended a different meaning.
Copyright. There must be meaningful human contribution to a copyrightable work.
Patent. A natural person must have significantly contributed to each claim in a patent application or patent.

USPTO’s recent guidance (Federal Register Notice 2024-02623; (89 FR 10043))

[ARS]

Human Creation Only -

Copyright

The Copyright Office has clarified that human creativity remains essential for works to be eligible for US copyright protection. Works entirely generated by AI w/ no meaningful human contribution are NOT copyrightable. (Ex: Monkey who took a selfie was not able to have copyright ownership over the photograph because the monkey wasn’t human)

If humans engage in selective input, arrangement, adaptation, or noticeable modifications of AI-generated output, you MAY be able to get copyright protection over the human contributions.

There is currently no new legislation pending or recommended to address copyright protection and AI-generated materials. Note: Change in leadership of the Copyright Office may impact their stance/policy on this.

Patents

AI can’t be inventor

Keep clear lines of what inventors contribute vs. how AI is used or is integrated into the invention process

Don’t plug inventions in publicly available AI - could be considered a public disclosure

24 of 44

Copyright�Who Can Create It?

Human Creation Only: Human creativity is essential for copyright registration eligibility.

AI Cannot be an Author: Works created by AI with no meaningful human input cannot be copyrighted.

Gray Area: If AI-generated output is modified enough, MAY be able to get copyright protection for human contributions

25 of 44

Patents�Who Can Invent It?

Human contribution must be significant: Human inventiveness is essential for patent eligibility.

AI Cannot be an Inventor: Inventions conceived of by AI with no meaningful human input cannot be patented.

Gray Area: If significant, inventive human contribution is mixed or embellished with AI-generated material, the invention SHOULD be patentable

26 of 44

Overview

IP as constitutional rights
AI versus Human Creation of IP
Common Legal Pitfalls for Higher Ed with AI use
AI Impacts on Specific Educational Areas
Best Practices for Higher Ed

27 of 44

Inventorship ≠ Ownership

Inventorship. A person who has contributed to conception of the invention, to the point that their idea is clear enough to reduce to practice.

Ownership. An entity with the authority to file patent apps and enjoys all the rights and benefits granted by the patent, if the patent is assigned to the entity.

[ARS]

I mentioned it on a prior last slide, but an important distinction to understand early on and through the course of this talk is inventorship vs ownership. They end up being the source of many of the disputes you could face.

The owner of a patent enjoys all of the rights and benefits granted by a patent – including the significant commercial rights to exclude others from making, using, selling, offering for sale, or importing the claimed invention. In the US, the owner is the entity who has authority to file patent applications.

An inventor, on the other hand, is a person who has contributed to conception of the invention, to the point their idea is clear enough to reduce to practice – put another way, clear enough enable someone skilled in the field to implement the invention in a practical form. A person qualifies as an inventor if they helped conceive something that’s described in at least one of the patent’s claims.

In the US especially, patents are granted to inventors and not directly to companies. In the constitutionally expressed interest of protecting inventors and the conception of their ideas, the persons who conceive of the idea are the ones who by default become the patent holders for the invention. Your business can only claim ownership of an invention if the inventor (whether this be an employee or contractor) assigns ownership to your business. We’ll talk later about HOW this is done, but just be aware that things like who conceived the idea in patenting and who authored the work in copyright establish implicit ownership by default.

It is worth noting, especially in the context of working with an engineering firm, that the inventor is not necessarily a person who reduces the invention to practice. So if someone strictly builds, code, manufactures, etc the invention under someone else’s direction who conceived the entire novel idea, the person or persons who implement are not inventors and claim no ownership unless they make additional inventive contributions during implementation.

Employment agreements

Sneaks into copyrights in sneaky ways with contractors, as well

28 of 44

IP Ownership

Creator is the owner unless contract says otherwise

If AI is the creator and did not sign a contract, then who owns the IP?

Require partners and employees to participate in IP requirements
Copyright. include “Work-for-Hire” provision
Patents. include “hereby assigns” provision

29 of 44

Whoever “writes” the code, owns the code

If AI writes the code, can the college own it?

Public repo warning. Keep code for patentable inventions out of public repositories.

Read Open-Source Software (OSS) license implications

Also applies to code adopted by the contractors.

Software Code Ownership

[ARS]

NDA ensures full confidentiality of all IP assets, trade secrets, and information you need to share with the software development partner for estimation and development purposes

Avoid IP Rights Traps. If I pay for development, will I own the code? The answer isn’t as simple as you’d hope. The software development agreement should explicitly provide for your ownership of any intellectual property that is created during the execution of the contract. Absent express language conveying an assignment, an SDA does not imply a transfer of patent rights or ownership of developed code.

Big copyright gotcha. Under copyright law, literally, the person who puts their fingers on the keyboard and types out a line of software is the “author” and owns the copyright. There’s a “work-made-for-hire” rule that makes it so the business that hires the employee who typed the code owns the code. But this breaks down when you’re hiring a contractor. The third category is the gotcha. A commissioned and copyrightable work will only be considered “work-made-for-hire” owned by you, so long as there is a written agreement signed by the developer that explicitly states that the work is “work-made-for-hire.” This has blocked acquisitions when purchasing companies have discovered that the company they’re trying to buy doesn’t actually control its own code.

30 of 44

Copyright Infringement

Copyright infringement occurs when one or more of an owner's exclusive rights are violated. (Not knowing is not a defense).
AI Risk. AI pulls from wide data set, including copyrighted content.
Fair Use. Exception/Defense for some uses of copyrighted content.
Recent Case Law. Showcase copyright implications of AI use as law develops.

31 of 44

Privacy and Disclosure Concerns

Exposure of sensitive student or university data

Disclosure risk for Intellectual Property

32 of 44

Classroom AI Use Concerns

Bias in grading

Opaque use of AI-based scoring

Misuse of AI generated Content

Over Reliance on AI detectors

Lack of access to AI tools

[ARS]

Bias: AI model may be biased for particular historical, cultural, or linguist styles; AI graders can’t reliably account for effort, growth, or process; If the AI is trained on past grading data from an instructor or institution, it can inherit existing biases.

AI-based scoring: If the instructor provides vague grading prompts, the AI may misinterpret the rubric or focus disproportionately on grammar, vocabulary, or structure instead of substance.

Overreliance on AI detectors

1. AI Text Detectors

These tools analyze writing samples to estimate whether they were produced by an AI model.

Turnitin AI Detection – Integrated into Turnitin’s plagiarism platform; gives a probability score on whether text was AI-written.

GPTZero – One of the first widely adopted detectors; checks for “perplexity” (how predictable the text is) and “burstiness” (variation in sentence structure).

Copyleaks AI Content Detector – Used by many universities; can detect AI-generated text in multiple languages.

Originality.AI – Paid tool, marketed for education and publishing; gives detection scores.

Sapling AI Detector – Focuses on professional/educational writing; free with limitations.

⚠️ Limitations: Accuracy varies, and false positives/negatives are common—especially with short text, mixed AI-human writing, or paraphrased AI output.

2. Plagiarism Checkers with AI Detection

Some plagiarism detection companies have added AI-detection capabilities:

Turnitin – As noted, now includes AI writing detection alongside plagiarism matching.

Unicheck – Combines similarity detection with emerging AI-detection features.

Copyscape / iThenticate – Primarily plagiarism-focused but evolving toward AI-awareness.�

3. Embedded LMS Tools

Some Learning Management Systems (LMS) integrate AI-detection into assignment workflows.

Canvas + Turnitin integration

Blackboard + SafeAssign with AI features� These make detection seamless for instructors inside their grading environment.�

4. Alternative / Indirect Detection Methods

Keystroke analytics (e.g., TypingDNA, Draftback in Google Docs) – Track how writing develops over time to verify authorship.

Contract cheating detection platforms (e.g., Authorship Investigations) – Look for style mismatches across multiple submissions.

Oral defenses / reflective assignments – Not a tool per se, but increasingly used as AI-resistant assessment strategies.

�

33 of 44

General AI Use Concerns

AI-Training Practices

Difficult to discern originality/novelty

Authorship & Inventorship Risks

Infringement Risks

34 of 44

Overview

IP as constitutional rights
AI versus Human Creation of IP
Common Legal Pitfalls for Higher Ed with AI use
AI Impacts on Specific Educational Areas
Best Practices for Higher Ed

35 of 44

AI Impacts on Audio Tech

PROS	CONS
AI-assisted mixing/mastering Voice synthesis & cloning Real-time audio analysis and feedback	Ethical & legal issues around AI voice replication Reduced emphasis on ear training Job displacement fears in production roles

36 of 44

AI Impacts on Fine Arts

PROS	CONS
New creative mediums Exploration of human/machine collaboration Broader accessibility to visual art.	Debates over authorship/creativity Copyright concerns with training data Potential devaluation of traditional techniques.

37 of 44

AI Impacts on Graphic Design

PROS	CONS
Speed in prototyping and mockups Smart layout and brand consistency tools Design idea generation	Risk of homogenized output IP issues with trained data Perceived erosion of the designer’s creative role.

38 of 44

AI Impacts on Library Sciences

PROS	CONS
AI-powered cataloging and metadata tagging Chatbots for research assistance Prediction acquisition and demand forecasting	Concerns about data bias in search AI Privacy issues with use data Over-reliance on automation for research guidance.

39 of 44

AI Impacts on Drones and Hardware Tech

PROS	CONS
Increased capability and analysis Customizable training Increased use cases	Privacy concerns Loss of control Data ownership questions Difficulty in securing patent protection of underlying algorithms

40 of 44

AI Impacts on Software Innovations

PROS	CONS
Faster code generation Increased capability and analysis Customizable training Increased use cases	Code ownership questions Difficulty in securing patent protection of software methods Debugging difficulties

41 of 44

Overview

IP as constitutional rights
AI versus Human Creation of IP
Common Legal Pitfalls for Higher Ed with AI use
AI Impacts on Specific Educational Areas
Best Practices for Higher Ed

42 of 44

Best Practices for Higher Ed

Institution-wide general guidelines

Institution-wide AI Use in IP guidelines

Course-Level Flexibility

Transparency

43 of 44

Best Practices for Higher Ed

Update Academic Integrity Policies

Integrate AI Literacy into Curriculum

Design assessments that discourage misuse

Ensure compliance with external funding sources & Partners

Develop NMC-specific, sandboxed AI models

[ARS]

Update Academic Integrity Policies

Include AI-generated content as a category of misconduct procedures with clear consequences (perhaps similar to how plagiarism is handled).

Integrate AI Literacy into Curriculum

Teach Students How AI Works (ethics, limitations, responsible use)

Critical Thinking Over Output (evaluate student judgment in using AI)

Prompt writing and evaluation (encourage students to critique both AI-generated and human writing to develop discernment)

Design Assessments that Discourage Misuse

Create opportunities/projects - to use AI responsibly but then also projects without it

Multi-stage projects (drafts, reflections, peer review, process documentation)

Oral defenses/in-person follow-up (confirm students understanding of their submitted work)

In-class assignments w/out access to AI (use to establish writing/thinking baselines)

Personalized or contextual questions (design assignments that are hard for AI to generate without significant human input)

Ensure Compliance with External Funding Source & Partners

Sponsors (e.g., NIH, NSF, DOD) may have emerging expectations around ethical AI use, IP ownership, and data handling—policies must align accordingly.

Develop NMC specific, sandboxed AI models

44 of 44

Questions or Comments? Contact us!

ashley@aurorapatents.com

(650) 380-6913

aurorapatents.com

@AuroraPatents

Podcast with us @ patentlystrategic.buzzsprout.com

mallory@breathe.law

(231) 268-4296

Breathe.law

@breathebrandprotection