GenAI in Communications
The world of AI technology is changing rapidly. This set of guiding principles and best practices is designed specifically for marketing, communications, public relations and public affairs staff at the University of Alaska, or any staff who fulfill those functions as part of their job duties. Informed by robust research, collaboration and real-world use cases, these best practices will help staff make sound judgment calls about AI usage and navigate the changing AI landscape with ethics, integrity and authenticity.
Philosophy: What we believe
AI is assistive, and not the creator of content. Humans remain responsible for every finished product.
Guiding principles
Authenticity - Authenticity is a pillar of UA’s identity. AI can serve as an assistant, but should not play a stand-alone role in our work. All interactions with AI should produce authentic materials. Inauthentic communication damages our shared identity and undermines our efforts at building trust with our community and our students.
Discernment - Critical thinking and careful evaluation can allow for thoughtful innovation, yet AI models require strict human oversight and judgment to guard against dangerous results. This requires accuracy, appropriateness, contextual understanding and adherence to guidelines. It is our responsibility to implement safeguards and monitor outcomes.
Accountability - Every individual, and ultimately the institution, is responsible for what we publish. Utilizing AI can result in unintended plagiarism and copyright infringement. Human expertise is required to properly manage these tools as they evolve. AI use should be consistent with all UA Board of Regents’ policies and OIT policies, as well as all university regulations and codes of conduct.
Transparency - AI use necessitates clear disclosure. All published content should include the proper attribution to the best of your knowledge. UA staff should maintain and value the connection to source materials.
Respect - We respect knowledge, creativity, the environment and our educational mission. As an institution of higher education, we must balance innovation with our responsibility to preserve learning, critical thinking and trust.
Examples of acceptable uses of Gen AI
The following examples reflect ways communicators across the UA system may utilize AI safely. This is not an exhaustive list. AI tools change quickly, and new use cases emerge regularly. Use these as a framework for thinking about appropriate use, not as a complete inventory of what's appropriate
AI can be a starting point for communications such as newsletters, social posts and event announcements. This could mean AI assists in building an example, an outline or providing feedback on human-crafted content.
AI can help generate ideas, headline options, content angles and structural approaches. These outputs should not be finished products. A user could ask AI what content other post-secondary institutions are publishing on a particular topic and use the examples to facilitate their own unique take.
AI can summarize long documents, organize research or sort data. Users should verify outputs before they rely on them. Please see OIT guidelines for information on how to evaluate data privacy, compliance, security concerns and risk.
AI transcription tools can be a practical option for converting audio to text. AI voice synthesis tools can convert written, staff-authored content into an audio format. Each user is responsible for the accuracy of all content.
AI can be a useful resource for learning on the go, with caution. Possible uses include asking AI to explain how something works, to walk through a process or to demonstrate a shortcut. Examples would be asking about formatting functions in Excel or for feedback on a list of questions for an upcoming interview. Users should stay within their skill level and only assign AI tasks they're capable of supervising and evaluating. If a user can't identify a mistake, they should not offload the task. Never trust outputs completely, as AI models are known to .
AI can generate draft alt text, captions and other accessibility elements.
AI upscaling tools can improve the technical quality of existing university-owned images. This can be appropriate when no misleading content is introduced and a human reviews the output. Some photo editing is in line with safe industry practice (such as a small extension in a sky background), but should only be used on benign components.
AI can assist with keyword research, readability analysis and metadata drafts for web content, such as asking AI to do a competitor analysis and provide helpful keywords that might boost visibility.
AI tools may be used for file organization, formatting, data sorting and similar behind-the-scenes tasks.
AI can serve as a useful thinking partner when preparing high-stakes or public-facing communications. An example could be asking AI to act as a journalist and anticipate potential questions, or to identify how stakeholders may interpret and respond to information.
As with all AI-assisted work, staff should exercise judgment about what content is appropriate to share. Confidential information, embargoed content, or details that could create institutional risk should not be entered into an AI platform, even in a strategic review context
Unacceptable uses of Gen AI
AI-generated visuals or audio should not appear in final university communications (including social media, promotional materials, websites or advertising) or be used to represent the university without explicit approval by the university or system office Marketing/Communications/Public Affairs director.
AI should not be used to produce quotes, bios, summaries or any content attributed to a specific person without verification. AI frequently fabricates details about individuals.
Use only UA-approved tools for work involving institutional data. Some consumer AI tools offer no data privacy or compliance assurances under UA's agreements. FERPA, HIPAA and other compliance obligations apply regardless of which approved tool you use. See UA's GenAI guidance at .
If you want to use an AI tool not already approved by UA, run it through OIT's evaluation process before using it for university work. OIT maintains a GenAI Tool Risk Assessment and security standards to guide use of new tools. See for more information.
No AI-generated text, image, audio or other content should be published or distributed without substantive human review. AI output is a draft, not a deliverable.
Media platforms may enable AI features automatically. This includes auto-publishing, auto-responding and AI-generated content running without human review. Monitor platform settings and opt out of unwanted features and exclude AI-generated content. Using a student's likeness in AI-generated content without consent is not acceptable.
AI meeting tools should not be active without clear notice to all participants. AI-generated meeting summaries sent via email become part of the public record and may be subject to FOIA.
A note on disclosure
When AI has played a meaningful role in producing a final communications product, staff should be prepared to disclose that use internally. The line is between AI as a tool (editing, formatting, brainstorming) and AI as a primary author. External disclosure standards are still being developed and will be addressed in future guidance.
Example scenarios
A communicator needs to draft a newsletter feature about a campus program
- Acceptable practice: Uses AI to generate an outline from interview notes, rewrites from their own voice, fact-checks against source material and submits for editorial review.
- Unacceptable practice: Accepts the AI draft with minimal editing, skips fact-checking and publishes.
A designer needs a campus photo prepared for a large-format display, but the original is too low-resolution.
- Acceptable practice: Uses an AI upscaling tool to enhance the existing image, reviews the output for distortions and uses it in the final design. No synthetic content introduced.
- Unacceptable practice: Uses an AI image generator to create a photorealistic image of students on campus because no suitable photo exists.
A communicator needs to post about a faculty member's research award across multiple platforms.
- Acceptable practice: Uses AI to adapt a press release into platform-specific posts for Instagram, LinkedIn and Facebook, then edits each for accuracy and tone before posting.
- Unacceptable practice: Posts the AI-generated versions without review.
A communications officer is preparing remarks for university leadership ahead of a high-profile announcement expected to receive media coverage.
- Acceptable practice: Shares the draft speech with an AI tool and asks how the remarks might land with different audiences, what questions journalists are likely to ask and whether any language could be mischaracterized. Uses the feedback to sharpen messaging before the speech is finalized. No confidential or embargoed information beyond the draft itself is shared.
- Unacceptable practice: Shares a draft that includes confidential budget figures and personnel decisions not yet made public. Relies heavily on AI sentence structure, resulting in remarks that sound inauthentic. The sensitive details entered into the platform are now outside the university's control.
UA Systemwide AI in Communications Working Group
These guidelines were developed by a cross-functional working group of eight communications professionals representing each of UA's three universities, the System Office, and affiliated units. The group convened in January 2026 and finalized the guidelines in June 2026.
Working group members:
- Rebecca Lawhorne, UA System Office of Public Affairs
- Maggie Demers, UA Foundation
- Vicki Daniels, »¨½·Ö±²¥Geophysical Institute
- Jeff Richardson, »¨½·Ö±²¥College of Fisheries and Ocean Sciences
- Katie Bender, UAA
- Summer Lazenby, Kenai Peninsula College
- Julie Stricker, »¨½·Ö±²¥Institute of Agriculture, Natural Resources and Extension
- Katy Jordan, UAS
The UA System Office of Public Affairs maintains these guidelines and updates them as the generative AI landscape evolves. Questions may be directed to ralawhorne@alaska.edu.