Johns Hopkins University

Effective October 11, 2017 (KSAS and WSE Homewood, WSE EP) Effective August 29, 2018 (KSAS AAP) I. SCOPE The following policy applies to all graduate students enrolled in full-time, part-time, or non-degree (visiting/special) programs in the Krieger School of Arts and Sciences (‘KSAS’) and the Whiting School of Engineering (‘WSE’), including Engineering for Professionals (“EP”) and KSAS Advanced Academic Programs (“AAP”).

证据定位: physical page 1; lines 4-10; Graduate Academic Misconduct Policy, Version Header and Section I Scope (Page 1)

If it is a first offense, and after consultation with the AIO, the faculty member/program chair/program director of graduate study and the student may reach an informal resolution of the alleged misconduct provided that: 1. It is a first offense, AND 2. The imposed sanction would be no greater than failure in a course, AND 4  3. The student and the faculty member/program chair/program director of graduate study agree to the terms of resolution of the alleged misconduct. (Neither the faculty member nor the student are obligated to resolve a compl

证据定位: physical page 4; lines 173-193; Coordinator-extended complete source context; excerpt offset 0–700

aint informally.) If such an agreement is reached, the the faculty member/program chair/program director of graduate study and student must complete the ‘Report of Informal Resolution Form’ (“Report”) and the the faculty member/program chair/program director of graduate study must provide a copy of that Report to their AIO (or designee) to be filed in the Office of Academic Affairs’ academic misconduct files. The Report is a part of the student’s academic conduct record. Unless required by law, regulation or contract, KSAS and WSE will neither confirm nor deny the existence of any informal resolution reached under this Section and will not disclose the contents of any such resolution to thi

证据定位: physical page 4; lines 173-193; Coordinator-extended complete source context; excerpt offset 700–1400

rd parties, but the information will be made available to any university panel in the adjudication of any future infractions.

证据定位: physical page 4; lines 173-193; Coordinator-extended complete source context; excerpt offset 1400–1525

A "preponderance of the evidence" standard will be used to determine responsibility for alleged violations of this policy. A "preponderance of the evidence" standard is an evidentiary standard that means "more likely than not." This standard is met if the proposition is more likely to be true than not true.

证据定位: physical page 6; lines 241-244; Graduate Academic Misconduct Policy, Section IV.D, Evidentiary Standard (Page 6)

Except in the case of an informal resolution for first time offenses with a faculty member, the student may appeal a panel’s finding of responsibility and/or sanction(s). A student must file their appeal within ten (10) university business days of the date of the notice of outcome on one or more of the following grounds: 9  • procedural error that could have materially affected the determination of responsibility or sanction(s); • new information that was not available at the time of the hearing and that could reasonably ha

证据定位: physical page 9; lines 416-443; Coordinator-extended complete source context; excerpt offset 0–700

ve affected the determination of responsibility or sanction(s); and • excessiveness of the sanction(s). Any appeal must be filed in writing with the AIO (or designee), who will review and send to the Cognizant Dean for a decision. An appeal will involve a review of the file by the Cognizant Dean; the appeal does not involve another hearing. On review of the appeal, the Cognizant Dean may seek additional information or clarification from the AIO, Hearing Panel, any relevant faculty member, the accused student, or any other relevant University or non-university personnel. On review of the appeal, the Cognizant Dean or designee may: • enter a revised de

证据定位: physical page 9; lines 416-443; Coordinator-extended complete source context; excerpt offset 700–1400

termination of responsibility and/or revise sanction(s); or • remand the matter to the original panel to reconsider the determination of responsibility and/or sanction(s); or • convene a new panel to consider the case. If the Cognizant Dean (or designee) has made a final determination with no panel remand/re- hearing, a final appeal determination will simultaneously be sent to the AIO, department chair, the faculty reporter, and the student. No further appeals are permitted.

证据定位: physical page 9; lines 416-443; Coordinator-extended complete source context; excerpt offset 1400–1930

Suspected violations of the Graduate Academic Misconduct Policy, including misuse of Generative AI, should be reported to the EP Academic Integrity Officer, Heather Stewart (heather.stewart@jhu.edu). Faculty should not independently resolve potential misconduct cases.

证据定位: lines 64-64; Coordinator-extended complete source context

Faculty should also consult local divisional guidelines for discipline-specific information if published.

证据定位: Generative AI Tool Guidance, table of contents

This page was developed collectively by the Johns Hopkins centers for teaching and learning to provide guidance on teaching strategies as they relate to or are impacted by generative artificial intelligence (AI).

证据定位: Background

One prominent use case is in content creation and course development. These tools can assist instructional designers and educators in generating engaging and interactive course materials, ranging from automated quizzes and assessments to case studies and customized learning modules.

证据定位: What Are Generative AI Tools?

🚫 Red: GenAI use is not permitted GenAI may not be used in any form. Students must complete all work independently without GenAI assistance, including drafting, editing, or revision.

证据定位: lines 38-40; EP Faculty Guidance, Course-Level Color Classifications, Red

⚠️ Yellow: Moderate GenAI use is permitted, with restrictions GenAI may be used in limited ways to support learning based on course policy. All GenAI use must be disclosed according to course expectations.

证据定位: lines 42-44; EP Faculty Guidance, Course-Level Color Classifications, Yellow

✅ Green: Liberal GenAI use is permitted, with expectations for transparency and accountability GenAI may be used throughout the workflow. Disclosure of GenAI use may be required, depending on course or assignment guidelines.

证据定位: lines 45-47; EP Faculty Guidance, Course-Level Color Classifications, Green

course must have a course-level color. Yellow-designated courses must also include assignment-level labels (i.e., colors) for all assessments and activities. Red- and green-designated courses need not apply categories at the assignment level.

证据定位: lines 50-50; EP Faculty Guidance, Implementation Guidance, Course Classification Requirement

Students must complete all work independently without GenAI assistance, including drafting, editing, or revision, unless explicitly stated otherwise.

证据定位: Student guidance: Generative AI Use Categories / Red; retained web-extracted text. Direct browser/HTTP returned 403; independently readable web extraction retained.

Disclosure of Generative AI Use Your instructor will outline specific standards regarding the level of GenAI-use disclosure required. Disclosure may include, but is not limited to: Full transcripts of chats or chat history Citations of a specific model or website used for particular tasks A short description of how GenAI was used A single citation per assignment, citing each model or website used Visual identification (for example, highlighting) of GenAI-generated content Failure to disclose GenAI use when required may constitute academic misconduct.

证据定位: lines 60-67; EP Student Guidance, Disclosure of Generative AI Use Header

Students remain fully responsible for verifying accuracy, identifying bias, and ensuring the quality of final submissions.

证据定位: lines 50-50; EP Student Guidance, Student Accountability

Save GenAI outputs or interaction records if disclosure or verification is requested.

证据定位: lines 76-76; EP Student Guidance, Documentation & Record-Keeping

AI Detectors – Many faculty ask about AI Detectors. JHU continues to explore and evaluate AI Detection methodologies, but at this time does not recommend incorporating AI Detection into your student evaluations. The technology is not reliable enough and often reports a vague probability of AI use. The reports do not include clear evidence like a plagiarism report from Turnitin.

证据定位: lines 28-28; CTEI Teaching and AI, AI Detectors Caution

All submitted work is expected to represent your own reasoning, writing, derivations, and problem-solving process.This course will use all available methods and technologies to evaluate authenticity of student work, including:AI-detection software Stylometric and authorship analysis Mathematical reasoning consistency checks Version-history analysis Oral follow-up questioning when necessary Comparison against prior submissions and known AI-generated patterns

证据定位: lines 132-137; EP Course 605.203.81 Syllabus, Student Work Authenticity Evaluation Methods

Do not enter nonpublic proprietary Hopkins data—including clinical data, financial information, and business records—into third-party AI tools not approved by Johns Hopkins, in accordance with the University Policy on Acceptable Use and Security of Johns Hopkins Information Technology Resources and the University Code of Conduct.

证据定位: lines 10-10; IT Guidelines for Responsible Use of AI, Protect Confidential Data

FERPA Guidelines Instructors should consider FERPA guidelines before submitting student work to generative AI tools like chatbots (e.g., generating draft feedback on student work) or using tools like Zoom’s AI Companion. Proper de-identification under FERPA requires removal of all personally identifiable information, as well as a reasonable determination made by the institution that a student’s identity is not personally identifiable.

证据定位: lines 16-17; Teaching Guidelines, FERPA Guidelines Section

HIPAA Guidelines Instructors including faculty, staff, and trainees who present should also consider HIPAA implications before using generative AI tools in the context of healthcare education.

证据定位: lines 18-19; Teaching Guidelines, HIPAA Guidelines Section

Use or disclosure of health information should be made in the context of generative AI only with JH-approved vendors where appropriate agreements are in place that offer appropriate protections to the information submitted and generated.

证据定位: lines 19-19; Teaching Guidelines, Approved HIPAA Vendors with Institutional Agreements

HopGPT is approved for sensitive data, including PHI and PII. In accordance with Johns Hopkins University and Johns Hopkins Medicine policies, you must act in a manner consistent with appropriate use of PHI/PII and be a responsible steward of the data you enter into and collect from HopGPT.Note: IRB approval is required for non-clinical research cases and SIP approval required for any clinical research cases or uses involving PHI or PII data.

证据定位: lines 10-10; IT HopGPT, Safe for Sensitive Data & IRB/SIP Approval Rules

Do not enter nonpublic proprietary Hopkins data—including clinical data, financial information, and business records—into third-party AI tools not approved by Johns Hopkins, in accordance with the University Policy on Acceptable Use and Security of Johns Hopkins Information Technology Resources and the University Code of Conduct.

证据定位: lines 10-10; IT Guidelines for Responsible Use of AI, Protect Confidential Data

Information shared with AI tools not approved by Johns Hopkins using default settings is not private and could expose proprietary or sensitive information to unauthorized parties.

证据定位: lines 11-11; IT Guidelines for Responsible Use of AI, Protect Confidential Data

HopGPT is approved for sensitive data, including PHI and PII. In accordance with Johns Hopkins University and Johns Hopkins Medicine policies, you must act in a manner consistent with appropriate use of PHI/PII and be a responsible steward of the data you enter into and collect from HopGPT.Note: IRB approval is required for non-clinical research cases and SIP approval required for any clinical research cases or uses involving PHI or PII data.

证据定位: lines 10-10; IT HopGPT, Safe for Sensitive Data & IRB/SIP Approval Rules

ChatGPT Edu is now officially available for Johns Hopkins University full-time faculty and staff. If you are not a full-time faculty or staff member at the university, you will need to speak with your divisional technical teams to explore available options. Before submitting an account request, please take a moment and read the resources on this page: https://engineering.jhu.edu/cmts/jhu-chatgpt-edu-account-request-for-faculty-and-staff/

证据定位: Adjacent retained excerpts recombined byte-for-byte; original order and source URL/hash preserved. lines 32-36; CMTS AI Tools Guide, ChatGPT Edu Section (Full Paragraph); excerpt offset 0–700 | lines 32-36; CMTS AI Tools Guide, ChatGPT Edu Section (Full Paragraph); excerpt offset 700–926

JHU ChatGPT EDU Beyond the $8.50 base monthly cost for your account, you will be charged for actual usage. ChatGPT Edu is the enterprise version of ChatGPT but with advanced features and privacy controls specific to education. https://openai.com/chatgpt/education/ Models & limits: https://help.openai.com/en/articles/11165333-chatgpt-enterprise-models-limits After the form is submitted with your cost center, our team will send you an invite. You must login with your JHEDID@JH.EDU.

证据定位: Adjacent retained excerpts recombined byte-for-byte; original order and source URL/hash preserved. lines 32-36; CMTS AI Tools Guide, ChatGPT Edu Section (Full Paragraph); excerpt offset 0–700 | lines 32-36; CMTS AI Tools Guide, ChatGPT Edu Section (Full Paragraph); excerpt offset 700–926

HopGPT (formerly Hopkins AI Lab) offers members of the Johns Hopkins community secure and easy access to Large Language Models (LLMs) from industry leaders including OpenAI, Anthropic, and Meta. Developed by IT@JH, this secure environment mitigates security and privacy risks while ensuring that data remains protected.

证据定位: lines 5-5; IT HopGPT, Providing Secure Access to Multiple Generative AI Models

HopGPT is approved for sensitive data, including PHI and PII. In accordance with Johns Hopkins University and Johns Hopkins Medicine policies, you must act in a manner consistent with appropriate use of PHI/PII and be a responsible steward of the data you enter into and collect from HopGPT.Note: IRB approval is required for non-clinical research cases and SIP approval required for any clinical research cases or uses involving PHI or PII data.

证据定位: lines 10-10; IT HopGPT, Safe for Sensitive Data & IRB/SIP Approval Rules

Microsoft Copilot (Web Version) Copilot on the web is free, is not HIPAA-compliant, and is not connected to your M365 apps or data. Uses web-based data and the latest large language models (like GPT-5) to generate responses. Does not access your organization's internal data (emails, documents, meetings, etc.). Useful for general productivity, research, and task automation. Offers enterprise-grade security and compliance *During our testing we found that Microsoft Copilot worked effectively with uploaded documents such as Word and Excel files. Select Continue with Microsoft to login with your JHEDID and password. After signing in select Work You can now start chatting with CoPilot

证据定位: lines 21-30; CMTS AI Tools Guide, Microsoft Copilot (Web Version) Section

Microsoft 365 Copilot Copilot for M365 is available for $215/user/year. This version is HIPAA-compliant and connects with your M365 apps and data. Order from the Software Catalog. *You must get approval and provide a cost center (full time faculty and staff Only) Uses your organization's internal data (emails, chats, documents, calendar, etc.) via Microsoft Graph to provide context-aware, work-grounded responses. Includes advanced features like: Copilot in-app experiences (e.g., summarizing emails in Outlook, generating slides in PowerPoint). Copilot Studio for building custom AI agents. https://copilotstudio.microsoft.com

证据定位: lines 38-44; CMTS AI Tools Guide, Microsoft 365 Copilot Section; excerpt offset 0–631

More information on CoPilot: https://it.johnshopkins.edu/ai/microsoft-copilot-for-m365/

证据定位: line 45; CMTS AI Tools Guide, Microsoft 365 Copilot Section; excerpt offset 631–718

The following pages are guidelines and best practices concerning generative AI tools and teaching. They will continue to evolve over time based on changes in the technology and use cases in the Johns Hopkins community.

证据定位: Generative AI Tool Guidance, table of contents

Although the benefits of generative AI tools in higher education are promising, approaching their implementation with care is necessary. Ethical considerations, such as bias detection and mitigation, should be addressed to ensure fairness and inclusivity.

证据定位: What Are Generative AI Tools?

Authorship. Authors are accountable and legally responsible for the entirety of their work. Therefore, AI and Large Language Models (LLMs) tools do not qualify as authors and cannot be listed as such on any publication. The Press will not accept any work that is substantially written by an AI or LLM tool.

证据定位: physical page 1; lines 8-10; Hopkins Press Generative AI Policy for Authors, Authorship Section (Page 1)

Transparency. Authors must disclose their use of AI or LLMs in the work they are submitting for possible publication. This must be done in two ways. First, the use of AI or LLMs must be acknowledged by a general statement of such use at the beginning of an article, in the front matter of a book, or in another prominent place in the work. Second, any use of generative AI or LLMs in text, images, graphs, tables, or other parts of the work must be cited as would be done with any external source. Specific citation styles for generative AI can be found in the following style guides from MLA, Chicago, and APA.

证据定位: physical page 1; lines 12-17; Coordinator-extended complete source context

Responsibility and Accountability. The author is fully responsible for the entirety of their work. They must ensure that it is original, accurate, does not plagiarize other’s work, and appropriately cites and references others’ work, including any content that is generated by AI or LLMs. The author accepts full responsibility and is solely accountable for any liability that may ensue from the use of AI or LLMs.

证据定位: physical page 1; lines 19-22; Hopkins Press Generative AI Policy for Authors, Responsibility and Accountability (Page 1)