HOPKINS AI LAB

Thank you for using Hopkins AI Lab (“AI Lab”). AI Lab is an application created by Johns Hopkins (“us”) using open-source code that allows users to connect to various large language models (“LLMs”), such as OpenAI GPT, Meta Llama, and Anthropic Claude, in a more secure and integrated manner. By using AI Lab, you acknowledge and agree to these Terms of Use.

1. Purpose and Limitations.

AI Lab is designed to assist users with information discovery across a wide range of research, academic, and clinical topics. AI Lab has not been trained, tested, or validated to provide healthcare decisions or medical diagnoses, legal advice, employment-related decisions, or academic decisions. As such, AI Lab is not a substitute for human judgment and should never be used as the sole basis for decision-making.  Use AI Lab as a supportive tool that you complement with your own expertise, critical thinking, professional judgment, and additional research. Use in human subject research involving participant data requires IRB approval.

2. Output Accuracy.

Please note that AI-generated outputs may not always be accurate, complete, or up-to-date. In some cases, AI has been shown to hallucinate output (i.e., generate fictional, false, or misleading output). Output accuracy can also drift over time. Please exercise caution in your use of AI Lab’s responses. You are responsible for verifying the accuracy and sufficiency of any output before relying on such output and in exercising your independent professional judgment. Output will only be retained in AI Lab for a period of thirty (30) days.

3. Model’s Terms of Use.

AI Lab allows for the use of third party LLMs in a secure environment. In addition to the terms here, your use of AI Lab is subject to the pass-through Restrictions on Use applicable to each LLM, available here: https://ailabsupport.it.jh.edu/models/. You agree to review and comply with all applicable Restrictions on Use, which vary depending on the model you choose to use.

We wanted to highlight terms that may be particularly relevant to some or all of our community.

Anthropic

  • Any use in a High-Risk Use Case (i.e., in Legal; Healthcare; Insurance; Finance; Employment and Housing; Academic Testing, Accreditation, and Admissions; or Media or professional journalistic content), requires that you implement the following safety measure: 1) have a “human-in-the-loop” (i.e., a qualified professional review the content or decision prior to dissemination or finalization) and 2) disclose to individuals about whom the output is intended to advise that you used the Anthropic Model to inform your decisions or recommendations. 
  • You understand and acknowledge that factual assertions in outputs should not be relied upon without independently checking their accuracy, as they may be false, incomplete, misleading or not reflective of recent events or information.
  • You agree that you shall not, and must not attempt to: (a) access any Anthropic Model to build a competing product or service, including to train competing AI models, without the prior written consent of Anthropic; (b) reverse engineer or duplicate the Anthropic Models; or (c) support any third party’s attempt at any of conduct restricted in this sentence.
  • You may use the Anthropic Models only in the countries and regions Anthropic currently supports.
  • You may not make public statements related to use of the Anthropic Models without Anthropic’s prior written consent, including in research or publications. 

You agree that your use of any Anthropic Model must comply with Anthropic Acceptable Use Policy (the “AUP”).

Meta

  • You must not use any Meta Model output or results to improve any other large language model (excluding Meta Llama 3).  
  • In addition, if a Meta Model is used to create, train, fine tune or improve an AI model, which is then distributed or made available, you must include “Llama 3” at the beginning of any such AI model name. 
  • You may not make public statements related to use of the Meta Models without Meta’s prior written consent, including in research or publications.

You agree that your use of any Meta Model must comply with the Meta Llama 3 Community License Agreement and Acceptable Use Policy (the “Meta Terms”).

Azure OpenAI

  • When using Azure OpenAI models, you are restricted from using the models to make decisions without appropriate human oversight as part of an application that may have a consequential impact on any individual’s legal position, financial position, life opportunities, employment opportunities, or human rights, or may result in physical or psychological harm to an individual.
  • You also may not use Azure OpenAI models to generate output for the express purpose of creating synthetic training data to develop or train AI models or systems that have substantially similar functionality to a Microsoft AI service, such as Azure OpenAI. The restrictions in this section do not prevent the use of a Azure OpenAI models to generate output for use as an input to dynamic AI models or systems.

You agree that your use of any Azure OpenAI model must comply with the Microsoft Enterprise AI Services Code of Conduct, and in particular, the Usage restrictions (the “Microsoft Terms”).

4. Usage Data; Feedback.

Johns Hopkins may collect data on your usage of AI Lab to help us improve its functionality. In addition, we welcome your feedback. By providing feedback, you grant us a perpetual, royalty-free right and license to use, modify, and incorporate your feedback in any manner, including in connection with further AI Lab development, without any obligation, royalty, or other restriction based on intellectual property rights.

5. Miscellaneous.

  • AI Lab is provided “as is,” without any warranties. It may contain bugs, errors, and other issues that could affect its performance and functionality. Johns Hopkins and its affiliates are not liable for any damages or losses resulting from the use of AI Lab. Your use of AI Lab is at your own risk.
  • We reserve the right to modify or terminate the AI Lab at any time.
  • Maryland law will govern these Terms of Use, except for its conflicts of laws principles.

Last updated 4-8-25