Generative AI: Ethical, Regulatory, and Legal Implications

• ABA Formal Opinion 512: Generative Artificial Intelligence Tools
  • "To ensure clients are protected, lawyers using generative artificial intelligence tools must fully consider their applicable ethical obligations, including their duties to provide competent legal representation, to protect client information, to communicate with clients, to supervise their employees and agents, to advance only meritorious claims and contentions, to ensure candor toward the tribunal, and to charge reasonable fees."
• ABA Formal Opinion 517: Discrimination in the Jury Selection Process
  • "A lawyer may not follow a client’s directive or accept a jury consultant’s advice or AI software’s guidance to exercise peremptory challenges if the lawyer knows or reasonably should know that the conduct will constitute unlawful juror discrimination."
• State Bar of Michigan: Artificial Intelligence for Attorneys FAQ

Accuracy

• Hallucinations: incorrect facts, misrepresentations, or completely made-up information
• Insufficient training data: How up to date are the documents the AI is being trained on? Are there any major content gaps in that data?

Bias

• AI systems can unintentionally perpetuate or even amplify biases in the training datasets. Users must be vigilant when using AI tools to identify potential biases and take steps to mitigate them.
  • At this point, using AI in hiring processes poses a high risk. There have been several instances of AI tools showing bias when used in video interviews as well as reviewing other application documents.

Communication with Clients

• Informed Consent: It is your professional responsibility to inform your clients if AI tools will be used during the course of their legal representation. Be transparent about how these tools will be employed, what sort of client data (if any) will be entered into the system, and any potential impacts on the case's outcome.
• Data Security Assurance: Ensure your clients that their data will be handled securely, and explain the steps taken to protect privacy and maintain confidentiality.

Competency

• Understand the AI Tool: Familiarize yourself with the AI tools you are using. What datasets was it trained on? What sources is it pulling from? Does it collect user input and data? How accurate are the outputs? Are there specific tasks ab AI tool could help complete?
• Due Diligence: "Trust but verify". Always verify the accuracy of the AI-generated output. 
• ABA Rule 1.1 Competence: Maintaining Competence: requires lawyers to stay informed about new technologies and use them responsibly.

Confidentiality

• Internal AI Systems: Use caution when inputting client identifying information into in-house AI tools (i.e. AI systems that were developed internally and have internal IT protection in place).
• Policy Adherence: Review your firm's policy on using generative AI tools to ensure attorney-client privilege confidentiality.
• External AI Systems: Never put sensitive or privileged client information into an opensource AI system, like ChatGPT. 

Copyright / Intellectual Property

• Training Data and Copyright: Ask AI vendors what data their AI models were trained on. If copyrighted materials were included in the training data, there could be risks of inadvertently violating intellectual property rights.

  • Thomson Reuters Enterprise Centre GmbH et al. v. ROSS Intelligence Inc. (1:20-cv-00613) - U.S. District Court for the District of Delaware

• Inputting Copyrighted Material: Avoid putting copyrighted or proprietary materials into an AI system that uses inputs to retrain or generate outputs.
• AI-Generated Outputs: The Copyright Office has determined "that the outputs of generative AI can be protected by copyright only where a human author has determined sufficient expressive elements."

  • Thaler v. Perlmutter et al. (23-5233) - U.S. Court of Appeals for the District of Columbia Circuit

• Copyright and Artificial Intelligence (Parts 1, 2, 3) - On May 9, 2025, Part 3 was released as a pre-publication version. A final version will be published in the near future and no significant changes are anticipated.

Legislation Trackers

*These are third party trackers and may not be up to date.

White & Case: AI Watch: Global regulatory tracker

Orrick: U.S. AI Law Tracker

BCLP: US state-by-state AI legislation snapshot

National Conference of State Legislatures: Artificial Intelligence 2024 Legislation

GAIIN: The Global AI Initiatives Navigator (OECD.AI's tracker for global public AI policies and initiatives)

Executive Orders on Artificial Intelligence

Federal Rules of Evidence: Rules 707 & 901

Tracking Changes To AI Evidence Under Federal Rules (Apr 1, 2025) - Law360

• Committee on Rules of Practice and Procedure (Jun 10, 2025)
• Committee on Rules of Practice and Procedure (Jan 7, 2025)
• Advisory Committee on Evidence Rules (Nov 8, 2024)

Other Material

Regulating Artificial Intelligence: U.S. and International Approaches and Considerations for Congress

NY State Courts Release AI Rules For Judges, Staff - Law360 (Oct 10, 2025)

Skadden Arps: US Federal Regulation of AI Is Likely To Be Lighter, but States May Fill the Void (Jan 14, 2025)

U.S. General Services Administration: AI Guidance and Resources

Centers of Excellence: Artificial Intelligence

Public Law 117-207: Artificial Intelligence Training for the Acquisition Workforce Act (Oct 17, 2022)

GAO Artificial Intelligence: Generative AI Technologies and Their Commercial Applications (Jun 20, 2024)

The EU Artificial Intelligence Act

• High-Level Summary of the AI Act

Currently, there is no federal AI legislation or regulation. There are several states that have proposed or enacted legislation on AI and it is a practicing attorney's duty to be aware of them.
 

• Who is responsible when AI-generated content causes harm or fails to meet legal standards?
  • Unpacking Liability When AI Makes A Faulty Decision (Mar 2025) - Law360
• Are there any intellectual property concerns? Was the AI trained on copyrighted material? Are you inputting IP into the system?
• Who owns AI generated outputs?
• Will this AI tool solve the specific issue I have?
• Does my firm have a policy on how to use this AI tool?
• Will my inputs be stored securely and remain private?