That's what the front desk staff told me this morning when I asked if anyone actually reads that stack of informed consent forms they handed to me.

I had to ask, "What happens if someone questions them or refuses to sign these forms simply because they don't understand what they say?"

"We fire the patient. We don't have time for trouble." I felt their pain. They are overworked, and of course, they have patients like me asking questions.

As someone who navigates healthcare both as a patient with a chronic condition and as a Responsible AI Healthcare Strategist helping organizations implement AI governance frameworks, this moment revealed a crisis hiding in plain sight.

If healthcare systems view basic informed consent as a nuisance, something only "problem patients" care about, how will they possibly handle AI consent requirements?

The Evidence Is Stark
Research shows fewer than 75% of patients correctly understand what they consented to, even in clinical trials. For complex concepts, that number drops to 50%.

Yet, these same systems must now disclose when AI influences diagnosis, explain algorithmic decision-making, clarify accountability when systems fail, and provide opt-out options.
Survey data reveal the gap: 75% of patients don't trust AI in healthcare, and 80% are unaware whether their doctor is using it. When informed, 80% say disclosure would improve their comfort.

The FDA Has Spoken
The January 2025 FDA guidance isn't optional; it establishes clear expectations for transparency, bias mitigation, and lifecycle management of AI-enabled medical devices. The WHO's AI ethics principles emphasize informed consent as a foundational principle.

But regulations alone won't fix a culture where asking questions makes you a "problem patient."

What Responsible AI Consent Actually Requires:

✓ Proactively disclose AI use in plain language
✓ Explain the AI's specific role in each patient's care
✓ Clarify who remains accountable for decisions
✓ Provide meaningful opt-out options
✓ Welcome patient questions as essential to governance, not trouble

This isn't about perfect systems. It's about fundamental respect for patient autonomy.

The Questions That Matter

• Does your organization treat patients who ask about AI as informed participants or as problems?
• What happens when someone questions AI use in their care, genuine dialogue, or "we don't have time for trouble"?

The future of responsible AI in healthcare doesn't depend solely on better algorithms. It depends on whether organizations can transform consent culture from legal protection theater into genuine patient empowerment.

About Dan Noyes

Dan Noyes operates at the intersection of healthcare AI strategy and governance. After 25 years leading digital marketing strategy, he is transitioning his expertise to healthcare AI, driven by his experience as a chronic care patient and his commitment to ensuring AI serves all patients equitably. Dan holds AI certifications from Stanford, Wharton, and Google Cloud, grounding his strategic insights in comprehensive knowledge of AI governance frameworks, bias detection methodologies, and responsible AI principles. His work focuses on helping healthcare organizations implement AI systems that meet both regulatory requirements and ethical obligations—building governance structures that enable innovation while protecting patient safety and advancing health equity

Want help implementing responsible AI in your organization? Learn more about strategic advisory services at Viable Health AI

Why the Paper Fix Won't Save You

Here's what should worry every general counsel and CMIO reading this: hospitals have spent decades trying to solve the readability problem with better paper, and it still hasn't worked. A 2024 analysis of consent forms from 15 large academic medical centers found the median form required about three minutes to read at a Flesch-Kincaid grade level of 13.9 — college freshman, while the average American reads at an 8th-grade level npj Digital Medicine study on surgical consent readability. That gap is not a rounding error. It's the reason prior research cited in that same study found patients actually read the full form somewhere between 1% and 45% of the time npj Digital Medicine study on surgical consent readability.

Now layer AI disclosure on top of that broken foundation. If your patients already can't parse a standard surgical consent form, handing them a denser one that also explains algorithmic decision-making isn't transparency. It's theater with extra paperwork. The same research team used GPT-4 to simplify those 15 forms and got the reading level down to 8th grade without a physician reviewer or malpractice defense attorney flagging any loss of legal or clinical sufficiency npj Digital Medicine study on surgical consent readability. AI can be part of the fix, not just part of the problem — but only if a human expert signs off on every simplified version before it reaches a patient.

What ONC's HTI-1 Rule Actually Requires — and What It Doesn't

I hear hospital executives cite "the ONC rule" as if it settles the AI consent question. It doesn't, and you need to know exactly where the line sits. The HTI-1 Final Rule creates first-of-its-kind transparency requirements for AI and predictive algorithms embedded in certified health IT, and it reaches deep into the market: ONC-certified health IT supports care at more than 96% of hospitals and 78% of office-based physicians nationwide ONC HTI-1 Final Rule. The rule gives clinical users — not patients — a consistent baseline of information about the algorithms they're relying on, so they can assess fairness, validity, and safety before trusting an output ONC HTI-1 Final Rule.

Read that again: the rule is built for clinicians assessing tools, not for patients deciding whether to consent to their use. Your patient-facing consent obligations are coming from somewhere else — state law. Utah's Senate Bill 226, in effect since May 2025, requires health care entities to disclose AI use whenever a patient asks whether they're interacting with AI, and specifically in "high-risk" communications like test interpretations or diagnostic results Association of Health Care Journalists report on state AI health laws. California's Assembly Bill 3030, effective January 2025, requires any health facility using generative AI to draft patient communications about clinical information to include a disclaimer that the content was AI-generated, plus instructions for reaching a human clinician Association of Health Care Journalists report on state AI health laws. Texas, effective September 2025, requires providers using AI to recommend diagnosis or treatment to review that output for accuracy before it enters the chart Association of Health Care Journalists report on state AI health laws.

If your compliance team is tracking one federal rule and calling it done, you're already behind three state statutes with direct, patient-facing disclosure duties — and more than 250 AI-in-healthcare bills have been introduced across state legislatures as of mid-2025 Association of Health Care Journalists report on state AI health laws.

Disclosure Alone Doesn't Buy You Trust

Here's the finding that should reshape how you think about this problem. A 2026 patient survey from the Coalition for Health AI found 75% of respondents already report using AI in some part of their care, yet only 13% feel very comfortable with it, and 51% say AI actually makes them trust the healthcare system less, versus 12% who say it increases trust CHAI patient survey on health AI and transparency. Ninety-three percent report at least one concern about AI in their care CHAI patient survey on health AI and transparency.

The CHAI researchers make a point I want every hospital board to sit with: transparency about AI use is broadly expected, but disclosure alone is not enough to build trust, and disclosing AI use without also explaining oversight and accountability can actually reduce it CHAI patient survey on health AI and transparency. More than 80% of respondents said their trust would increase if clear accountability measures were visibly in place CHAI patient survey on health AI and transparency. A separate JAMA Network Open survey of 2,039 U.S. adults found that patients' trust that a health system will use AI responsibly tracks their general trust in that system — but has no measurable relationship to their health literacy or AI knowledge JAMA Network Open study on patient trust in AI. Patients who had previously experienced discrimination in a health care setting were significantly less likely to trust their system to use AI responsibly or protect them from AI-related harm JAMA Network Open study on patient trust in AI.

Put those two studies together and the implication is unavoidable: you cannot out-communicate a trust deficit rooted in your institution's broader relationship with patients. A better-written consent form helps. It is not a substitute for the accountability structure — named humans responsible for AI outcomes, a visible override process, documented bias testing — that actually earns belief.

A Practical Checklist for the Next Board Meeting

None of this requires waiting for a final federal rule. Here's where I'd start with a governance committee this quarter.

• Audit consent-form reading level before adding a word about AI — if you're above 8th grade now, more AI language only worsens comprehension.
• Separate ONC/HTI-1 compliance (clinician-facing algorithm transparency) from patient-facing consent obligations (state disclosure laws) — different projects, likely different owners.
• Map every state you operate in against the patchwork — Utah, California, Texas, and Illinois already impose different disclosure or restriction duties.
• Pair every AI disclosure with a visible accountability answer: who is responsible when the AI is wrong, and how a patient reaches a human to challenge it.
• Track disclosure as a trust metric, not a legal checkbox — measure whether AI disclosures raise or lower patient comfort, and adjust accordingly.

Continue the Conversation

If this resonated, here is where to go next: Join the AI-in-Healthcare Workshop · Get the Books · Contact Dan