Dartmouth Communications Office Used AI to Draft Social Media Posts
A former Dartmouth College communications employee created three AI projects on the institution's Claude enterprise portal to help draft and edit social media captions, according to documents reviewed by The Dartmouth student newspaper.
Micky Bedell, who recently left her role as assistant director of social media, uploaded templates and guidelines to Claude projects titled "Dartmouth Social Caption Writer," "Dartmouth Social Carousel Brainstorm," and "Dartmouth Social Caption Editor." The projects were last edited between March and April 2026.
The projects functioned as a "knowledge base" - a collection of prompts and files that other members of Dartmouth's enterprise Claude group could access to provide context to the chatbot. The enterprise group includes all campus community members: students, professors, and staff.
How the AI System Was Configured
Bedell's projects instructed Claude to mimic Dartmouth's brand voice by sounding "human and relatable" and "confident and proud, but not boastful." The system was told to write at an eighth-grade reading level and consider whether "someone's mom" would understand the captions.
The projects included an "AI Pitfalls Guide" that banned more than 100 words and phrases commonly found in AI-generated text. Prohibited terms included "embark," "robust," "in today's fast-paced world," and "in the ever-evolving landscape of."
Both projects flagged em-dashes and "emoji-led bullet points" as telltale signs of AI writing that should be avoided.
The templates also referenced President Sian Leah Beilock's strategic priorities, including "Dartmouth Dialogues: Fostering Brave Spaces," "Commitment to Care: Student Mental Health and Well-Being," and "The Dartmouth Climate Collaborative."
College Says AI Use Is Standard Practice
Janna Barnello, Dartmouth's spokesperson, said Bedell's departure had nothing to do with her AI work. "My former colleague got an exciting new job, and that is why she left Dartmouth," Barnello wrote in an email, referring to Bedell's new position as senior social media manager at King Arthur Baking Company.
Barnello characterized the projects as an "exploration" of whether Claude could help draft captions. "A number of staff across Dartmouth are experimenting with these tools in their work, and this is one example," she said.
All social media posts undergo review by at least two people before publication, regardless of whether they were written by staff or drafted with AI assistance, according to Barnello.
The Office of Communications maintains a voluntary internal working group focused on AI use, Barnello said. Members discuss where these tools fit responsibly into workflows and what training staff need to use them effectively.
"We're thoughtful about when these tools belong in our work and when they don't," Barnello wrote.
What This Means for Communications Professionals
The Dartmouth case illustrates a practical approach to AI in communications: building guardrails into AI systems to prevent common pitfalls while maintaining human oversight. The banned-words list and readability guidelines show one institution's method for keeping AI output aligned with brand standards.
For communications teams considering similar tools, the setup raises questions about governance, training, and where human judgment remains essential. Bedell's role on an "institutional AI Learning and Development Committee" suggests Dartmouth treated AI adoption as a skill that required structured learning.
If you manage social media or communications teams, understanding how to configure AI tools for your brand voice - and how to oversee their use - is becoming a standard competency. AI Learning Path for Social Media Managers covers frameworks for integrating these tools responsibly, while Claude AI Courses provide hands-on training with the specific platform Dartmouth used.
Your membership also unlocks:
