Executive Summary/ Key Takeaways

• What is the "Work Product Doctrine"?
  The work product doctrine is a legal principle that protects documents and materials (work products) prepared by or for an attorney in anticipation of litigation from being disclosed to the opposing party. While attorney-client privilege protects "communications" between a lawyer and client, the work product doctrine specifically shields the "strategies, mental impressions, and legal analyses" developed for a case.
• Does attorney-client privilege apply to documents generated by AI?
  Generally, no, if the documents were created using consumer-grade AI tools without attorney involvement. In United States v. Heppner, the court ruled that AI is not an attorney and the “chat” with AI was not aimed at obtaining legal advice, and that interactions with consumer AI platforms lack a "reasonable expectation of confidentiality" because their privacy policies often allow data collection and third-party disclosure.
• Can AI-generated documents be protected under the work product doctrine?
  If a user generates documents using AI based on their own judgment without specific instructions or supervision from an attorney, such documents are unlikely to be protected. The Heppner ruling suggests that the outcome might have been different if the AI tool had been functioning under the direct supervision and instruction of legal counsel.
• Can privilege be secured by sharing AI-generated documents with an attorney after they are created?
  No, privilege cannot be applied retroactively. The applicability of privilege is evaluated based on the circumstances and purpose at the "time of creation". If a document was generated in a non-confidential environment using public AI tools, subsequent delivery to an attorney will not cure the lack of privilege or shield it from discovery.
• In what specific M&A scenarios does AI usage pose a legal risk?
  Unlike business licenses, product licenses may be transferred under certain conditions. However, such transfer generally requires that no changes be made to the approved product particulars. If a partial change application is required, the timing of the transfer may affect the deal structure and the closing schedule of the transaction.
• What governance measures should companies implement for AI in M&A?
  Three key measures are essential:
  1. Involve counsel early: Ensure that any AI use for legal analysis is conducted under the direction of an attorney to establish a basis for privilege or work product protection.
  2. Use Enterprise AI: Use only enterprise-grade AI platforms that guarantee confidentiality and prohibit data training, and restrict the use of consumer-grade tools.
  3. Establish AI Governance Clauses: Explicitly define the conditions for AI use and the preservation of privilege in transaction documents such as NDAs and LOIs.