Practice Management Solutions infused with AI are filled with huge promise for legal professionals, but attorneys need to balance their excitement against an ethical framework to make sure that they are safeguarding their work-product and attorney-client privileged information.  New rules of ethics addressing the use of AI are being promulgated around the US, and here is a brief summary on some of the emerging trends:

Duty to use AI ethically and responsibly

1-An attorney has a duty to never submit confidential client info or privileged work product to a chatgpt model that uses it to train an LLM model.

2-You have a duty to understand how any 3d party AI you are using in your practice handles your data – and do not use AI where your clients confidential data and privileged work product is being used to train an AI model.

3-Attorneys have an obligation to understand any bias that may be embedded in the AI results, which includes a duty to understand the data set that is being used to generate AI results.

*When LawToolBox uses Azure AI to generate summaries the data set used to is always defined by the end-user. For example, LawToolBox AI can be used to analyze a URL or file that the end-user identifies, or it could be used to look through files in the end-user’s own Microsoft tenant, but without exposing the data or results to LawToolBox.

4-The ethical bar is higher for the use of Generative AI in contrast to the Semantic AI employed by companies like Lexis or Westlaw to search through millions of court opinions.

5-A person licensed to practice law must review all AI generated work (just like you would review the work of a prarlegal or new associate). This may seem obvious – but don’t blindly trust AI generated work product – it hallucinates and makes mistakes.

6-If an attorney uses AI to generate any portion of a legal document filed with the court they may have an obligation to disclose the extent to which AI was used to generate the document. Some jurisdictions go even further requiring the certification that Confidential Information was not disseminated. Some jurisdictions also require every section that was generated using Generative AI to be identified to the court.

7-Attorneys should include in client agreements information about whether you will use AI and how you will bill for it. You cannot bill 1 hour for work that AI did in 5 minutes, but you can bill for the amount of time you spent drafting and revising AI prompts. Flat rate agreements or contingent fee are one way to capture this value.

**LawToolBox Responsible AI Guarantee: LawToolBox uses Microsoft Azure AI and Microsoft Copilot to generate AI results which by design will never train on your data. Because LawToolBox AI operates completely within the secure and safe Microsoft container for Azure AI and Copilot, LawToolBox does not (and cannot) store any AI generated results, and none of your data is ever used to train a Microsoft or LawToolBox Large-Language-Model (LLM). LawToolBox AI and LawToolBox for Copilot stay within the privacy walls of your own MS secure data container