Your Chief AI Officer · Wednesday, October 28, 2026 · Act today
JAMA letters debate guardrails for advertising inside AI decision support tools
Opinion4 sourcesFor patients and employees
Letters in JAMA respond to a Singh and Mello Viewpoint warning that advertising in AI clinical decision support can influence care. One proposes a clinical query firewall; another compares free tools to past drug company giveaways. The authors reply that ending the practice altogether is unlikely.
Three letters published in JAMA continue a debate started by a Viewpoint from Drs Singh and Mello, which warned that advertising inside AI-powered clinical decision support (CDS) tools could influence care at the point of clinical decisions. The exchange matters to any organization whose clinicians reach for free AI tools during patient care, because the business model behind those tools now has a compliance and clinical governance dimension.
The first letter proposes a clinical query firewall as the most important guardrail. Its argument is that a CDS query is not a simple search term. It may encode a provisional diagnosis, treatment uncertainty, imaging or molecular findings, medication history, comorbidities and the clinician's uncertainty about a specific patient. Turning that query into an advertising signal, the writer says, converts clinical reasoning into commercial inventory. For privacy and legal teams, that framing puts query data in the same category as other sensitive clinical information.
A second letter, from Dr Ziegelstein, compares free AI CDS to the drug company sponsored lunches, branded pens and giveaways of the 1990s, noting that there was no free lunch then and there is no free CDS now. It cites the Viewpoint's observation that while some products use subscriptions, others such as OpenEvidence use advertising revenue to offer access without charge.
In their reply, Singh and Mello agree that ending the practice would be the simplest fix but call it unlikely, since advertising is potentially more lucrative than subscriptions. They see clinician choice of ad-free tools as more realistic, particularly if employers nudge clinicians that way. They point to history: once Epocrates, a free drug information app, began supporting ads at the point of care, its credibility with physicians eroded and services such as UpToDate ultimately won out.
The evidence here is limited. These are opinion letters, and no study of patient impact from ads in AI CDS is cited. What they offer is a clear governance question, not a measured harm.
The move is practical. The Chief AI Officer, Chief Medical Officer and compliance lead should inventory the free and ad-supported AI decision support tools clinicians actually use, learn how each handles query data, and set a policy on which are permitted. Where a sanctioned, ad-free option exists, making it the easy default protects clinical judgment, protects patient information, and spares clinicians from sorting out commercial influence on their own.
Do this week. Inventory free or ad-supported AI decision support tools clinicians use and set a policy on which are permitted.
Run this
The prompt to act on this item, for your own model.
Paste it into Claude, ChatGPT or Copilot and fill the bracketed slots with your own numbers. The output is a draft for your team to check, built only from the sources on this page. Never paste patient data. How these prompts are written.
You are an operations advisor to the executive team of [your organization], working with the Chief AI Officer, the Chief Medical Officer and the compliance lead. Three letters in JAMA continue a debate started by a Viewpoint from Drs Singh and Mello, which warned that advertising inside AI-powered clinical decision support tools could influence care at the point of clinical decisions. The first letter proposes a clinical query firewall. It argues that a query can encode a provisional diagnosis, medication history, comorbidities and the clinician's uncertainty about a specific patient, and that using it as an advertising signal turns clinical reasoning into commercial inventory. A second letter, from Dr Ziegelstein, compares free AI decision support to the drug company giveaways of the 1990s and notes that some products, such as OpenEvidence, use advertising revenue to offer free access. In their reply, Singh and Mello call ending the practice unlikely. They see clinician choice of ad-free tools, nudged by employers, as more realistic, and recall how Epocrates lost credibility with physicians after adding ads while UpToDate won out. These are opinion letters, and no study of patient impact is cited. Your task: help us inventory the free or ad-supported AI decision support tools our clinicians use and draft a policy on which are permitted, work we can finish this week. Produce a one-page memo for the executive team with these sections: Why this matters; Inventory, as a table listing each tool, its business model, how it handles query data, and what we must still verify; Proposed policy, sorting tools into permitted, restricted and not permitted, with a sanctioned default; Owners and next steps; Open questions. Use these details: [the AI decision support tools you know clinicians use], [your sanctioned ad-free option, if any], [the number of clinicians in scope], [your EHR]. Use only the facts above and the details I fill in. Cite the sources by name: the JAMA letters and the Singh and Mello reply. Flag anything uncertain or that needs verifying, especially vendor data practices. Include no patient data. Nothing here is medical advice. Say clearly that this is a draft for the team to check.
- The claim.
- The letter writers argue commercial interests can shape medical judgment through ad-supported AI tools, and that clinical queries need protection.
- The evidence.
- These are opinion letters in JAMA; no study of patient impact is cited.
By seat
For the Chief AI Officer. Add ad-supported tools and query data use to AI vendor review criteria.
For the Chief Medical Officer and physicians. Ask physicians which free AI search tools they use and whether they show sponsored content.
For the Compliance, privacy and legal. Review whether clinician queries containing patient details flow to advertisers under tool terms.
Checked before publishing. A primary source: a peer-reviewed journal or a regulator.
Sources
Drafted by Lumen from the sources above, scored by a second pass on six criteria, then checked by rule against those sources (every figure, two independent outlets or a primary source) and published only when it passes. A person can pull any item, and corrections are published here. Apolitical, always. Nothing here is medical advice. How Your Chief AI Officer is made.