The week HCIC turned 30, Johns Hopkins Bayview began mailing letters to 3,182 patients. An employee had uploaded a spreadsheet of their information to an AI tool the hospital never approved. Days earlier, the largest U.S. banks said their customers had been handing login credentials to AI agents nobody at the bank had vetted.
Recorded live at the Healthcare Interactive Conference in Orlando, Chris Boyer is joined by Danny Fell of Unlock Health and Jeff Steblea, immediate past president of NESHCo. Both have spent their careers in healthcare marketing, long enough to have watched every previous wave of technology arrive with the same promises. This conversation stays on a quieter problem. The door is already open, and the people walking through it picked their own AI.
The panel explores:
- Why employees route around approved AI tools, and why Harvard Business Review’s new cover story traces it to trust in the organization rather than trust in the technology
- What happens when the assistant someone chose feels as personal as a favorite team, and the company hands them a different one
- Why the banks’ “rogue agent” problem looks a lot like customers inviting agents in, and what that means for patient portals, scheduling and refills
- What a personal agent says when asked about its own security risks, and why the panel doubts anyone will follow that advice
- Where “a little too far” sits in an industry that already tolerates medication errors, and who gets to decide
If your staff and your patients have both picked their own AI, what is your AI policy actually governing?
MENTIONS FROM THE SHOW
- Healthcare’s Long Walk Toward the Patient, the free eBook marking 500 episodes, plugged by Reed in the open: https://www.touchpointpodcastbook.com
- Danny Fell on LinkedIn: https://www.linkedin.com/in/dannyfell/
- Unlock Health: https://unlockhealth.com/
- Jeff Steblea on LinkedIn: https://www.linkedin.com/in/jeffsteblea/
- WMAR-2 News, “Johns Hopkins alerts 3,000+ patients after data sent to AI tool,” October 2026: https://wmar2news.com/local/patients-data-sent-to-ai-tool-johns-hopkins-bayview-medical-center-oct-23-2026
- Harvard Business Review, “Why Even Your Best Employees Override AI Systems That Work,” by Das Narayandas: https://hbr.org/archive-toc/BR2606
- Axios, “Scoop: Unvetted AI agents are accessing U.S. bank accounts,” October 22 2026: https://www.axios.com/2026/10/22/ai-agents-banks-unauthorized-accounts-access
- PYMNTS on JPMorganChase, Citigroup, Bank of America and Charles Schwab all saying outside AI agents cannot move customer money: https://www.pymnts.com/?p=4326961
- OpenAI, Finances in ChatGPT: https://help.openai.com/en/articles/20001222-app-permissions-in-chatgpt
- Plaid, the account-linking service: https://plaid.com/
- Technology Magazine on the PHP vulnerabilities: https://technologymagazine.com/news/the-php-crisis-explained-uncovering-flaws-in-web-code
- TechRepublic on Meta’s Muse: https://www.techrepublic.com/article/news-meta-muse-ai-agent-us-launch/
- Fierce Healthcare on Utah’s AI prescription renewal pilot with Doctronic: https://www.fiercehealthcare.com/ai-and-machine-learning/deep-dive-doctronics-ai-prescription-refill-pilot-program-utah
- Nick Bostrom, “Ethical Issues in Advanced Artificial Intelligence,” 2003: https://nickbostrom.com/ethics/ai
- Chris Boyer on LinkedIn: https://www.linkedin.com/in/chrisboyer/
- Chris Boyer website: http://www.christopherboyer.com/
- Chris Boyer on BlueSky: https://bsky.app/profile/chrisboyer.bsky.social
- Reed Smith on LinkedIn: https://www.linkedin.com/in/reedtsmith/
- Reed Smith on BlueSky: https://bsky.app/profile/reedsmith.bsky.social
Recommendations from this episode:
- Jeff: Little Eyes by Samanta Schweblin: https://www.publishersweekly.com/9780525541363
- Danny: Trader Sam’s Grog Grotto: https://touringplans.com/blog/trader-sams-poly/
- Chris: Vote: https://vote.gov/

