Generative AI has raised a multitude of legal questions. It is also challenging established legal norms, necessitating the development of new legal regulations and frameworks.
The "Generative AI Legal Explainer" is an ongoing initiative aimed at providing individuals with the kind of responses that legal experts might offer informally (thanks to NYU Clinical Assistant Professor Jenny McPhee for sharing this). The resource offers commonly asked questions accompanied by a concise answer designed to provide a clear understanding of the most probable outcome in a typical scenario.
Below are some of the most frequently asked questions and a summary of responses based on real-life legal cases.
Use of Personal Content:
Can GenAI companies train their systems on images, text, code, or other things I’ve made without getting my permission?
Probably Yes. GenAI companies can train their systems on images, text, code, or other things you've made without obtaining your explicit permission, primarily due to the application of the fair use doctrine in US copyright law.
Ownership of AI-Generated Content:
Is GenAI allowed to produce outputs “in the style of” a human artist?
Probably Yes. GenAI is generally not prohibited from producing outputs "in the style of" a human artist as copyright primarily protects specific works, and style copying is challenging to address through traditional copyright law.
When I use GenAI, do I own any of the outputs?
Probably No. The ownership of GenAI outputs is currently in a state of flux, and it's primarily considered that the machines that create them, not the humans who prompt them, are the authors according to US Copyright Office decisions.
Data Collection by AI Products:
When I use GenAI products like ChatGPT, can they collect my data?
Data Removal from Training Datasets:
Can I remove my personal data from GenAI training datasets?
Probably Not. Currently there is no foolproof method to compel companies to disclose whether your personal data is in GenAI training datasets, and even if such data were identified, it is technically challenging to fully remove it from these datasets without unpredictable consequences for deep learning models.
Defamation by AI Tools:
Can I sue a GenAI company for defamation if its tool generates false information about me?
Probably Yes. You could potentially sue a GenAI company for defamation if its tool generates false information about you, but the success of the lawsuit would depend on a variety of factors, including whether the false information is presented as a statement of fact and whether the GenAI company was negligent or aware of the harm caused.
It is important to remember that these questions and answers should not serve as a substitute for legal counsel. For legal guidance on specific actions, consulting with an attorney remains crucial.
To delve deeper into the complexities of generative AI's legal landscape, read the full Knowing Machines "Generative AI Legal Explainer."
NOTE: Portions of this post were created with the assistance of Generative AI.