American University's Guide to Responsible AI Use: Principles, Policies, and Best Practices

American University's approach to Artificial Intelligence is deeply rooted in its core values, including a commitment to human dignity, equity, integrity, and free inquiry. The university champions truth-seeking, community building, and making a positive impact on the world. Central to this is the principle of **Principled Use**. This mandates transparency in AI usage; users must disclose when an AI tool has wholly or partially created a work product and, where appropriate, how it was used. Honesty about AI involvement is paramount, though the specific form of disclosure may vary by context. Furthermore, the **AI user bears the ultimate responsibility for validating the accuracy of any AI-generated content** using trusted, first-party sources. Continuous monitoring of the reliability of AI outputs is essential. Users are accountable for how they utilize AI-created content and the impact of that use on their work. This underscores that AI is a tool to augment human capabilities, not replace human judgment or responsibility.

The responsible use of AI hinges on robust data privacy and security measures. American University strictly advises users to **treat any information provided to open (non-enterprise) AI tools as if it were public.** This means refraining from entering any sensitive data, or data classified as Confidential or Official Use (as defined by the university's Data Classification Policy), into AI platforms unless explicitly approved by a university official. The Data Classification Policy offers detailed information on protected data types. Furthermore, users must not upload, input, or analyze university data using AI tools that have not been approved by the Office of Information Technology (OIT). Contracts for approved AI tools are vetted to ensure that university data is not used for model training or is isolated in secure environments inaccessible to external parties. Users must also be aware that most AI tools and services utilize user input and data to train their models. Data entered into these tools, including personal information, may not be deletable or retrievable later. Some advanced AI tools, often termed 'agentic AI,' may also access data from other open windows or applications on a user's device, necessitating heightened caution.

Researchers at American University are guided to maintain high standards of integrity and transparency when using AI in their work. **Transparency in AI usage is paramount.** Researchers must clearly track and, where appropriate, disclose when, how, and which AI tools (including specific versions) were employed, whether for data analysis, content generation, or writing assistance. Researchers are fundamentally responsible for the reproducibility of their research outcomes. **Ensuring the accuracy of AI-generated content is critical.** All AI outputs must be rigorously verified against trusted sources, especially for references and factual claims. Researchers bear full responsibility for the integrity of any material incorporated into their research. **Protecting data privacy** is also a key concern; sensitive or proprietary information should not be input into publicly available AI tools. Prioritizing self-hosted AI systems or, failing that, AI systems meeting strict security standards (e.g., ISO 27001, SOC 2) with opt-out options for data training is advised. When sharing data for reproducibility, efforts must be made to ensure human subjects cannot be re-identified. Researchers should **evaluate potential bias** in AI-generated outputs, using automated fairness checkers as supplementary tools. Consideration must be given to how results might affect vulnerable populations or perpetuate existing inequities. **Respecting intellectual property** involves appropriate attribution of AI-assisted contributions, without listing AI systems as co-authors, as they cannot be held accountable for research integrity. Adherence to publisher and tool-specific guidelines on attribution, licensing, and acceptable use is mandatory. AI should not be used for tasks requiring human ethical judgment, such as peer reviews or proposal evaluations. If AI assists in grant writing or compliance documentation, the principal investigator must verify and disclose its involvement. Finally, researchers must use AI responsibly and ethically, avoiding the generation or dissemination of malicious or misleading content, and being mindful of how input data might be reused for future model training.

Students at American University are responsible for understanding and adhering to the AI policies outlined in their course syllabi concerning the use of AI in each class. Misuse of AI or its use without proper permission or disclosure may be considered a violation of the Academic Integrity Code. This responsibility extends to any extracurricular submissions or outputs where AI tools are utilized. Students are accountable for the outcomes generated by any AI tools they employ. They must be mindful of how to best use and not use these tools, recognizing that any errors or inaccuracies in AI responses become their own mistakes, which they must control through review and correction. Students are partners in achieving course learning objectives and should not offload their learning entirely to AI tools. Given the inherent privacy concerns associated with AI tools, students should proactively seek permission before inputting the work of others into an AI tool, aligning with the AU Responsible Use of University Technology Resources Policy. Students who lack access to AI tools (paid or otherwise) should not be disadvantaged. In such cases, students are encouraged to discuss their ability to meet course requirements directly with their instructor, department, or the Office of the Dean of Students.

American University is committed to harnessing the power of Artificial Intelligence while upholding its core values and ensuring the responsible, ethical, and secure use of these transformative technologies. The comprehensive guidance provided—covering principled use, transparency, human oversight, data privacy, and specific directives for teaching, research, staff, and students—serves as a roadmap for navigating the complexities of AI integration. By fostering AI literacy, demanding accountability, and prioritizing the protection of sensitive information, the university aims to empower its community to leverage AI effectively and ethically. This proactive approach ensures that AI serves as a tool to enhance learning, research, and operations, rather than posing a risk. As AI continues to evolve, American University will remain diligent in updating its policies and practices to reflect the latest advancements and best practices, ensuring a future where AI is used responsibly to benefit the entire community and beyond.