When a cancer diagnosis lands, the mind often moves faster than the body can catch up. We lie awake running probability we never asked to learn, and the emotional load becomes its own kind of treatment. According to a recent commentary in The Cancer Letter, the mental health care that follows — psycho-oncology — is quietly being reshaped by artificial intelligence. The piece argues, plainly and without theatrics, that the ethical scaffolding to keep that shift safe has not yet caught up with the technology itself.
The care we have, the framework we don't
The commentary makes a pointed case: AI tools are already being woven into how clinicians screen for distress, triage support, and offer cognitive interventions for people navigating cancer. But without a shared ethical compass — one built jointly by developers and mental health providers — we risk accelerating access to care in ways that outpace our ability to protect the people receiving it.
This is the kind of gap that doesn't announce itself in a clinic visit. It shows up later, in the quiet worry about whether an algorithm truly saw you, or whether your most vulnerable disclosures were handled with the weight they deserve. For those of us sitting with patients, or sitting as patients, the question shifts from "does this work?" to "does this work for me, in a way I can trust?"
What "patient-centered" actually asks of us
The Cancer Letter frames the goal in four words: safe, equitable, trustworthy, and patient-centered care. They sound simple until we sit with what each one demands. Safe means we know where our disclosures travel and who reads them. Equitable means the same quality of support reaches the person in a rural oncology ward and the one in a flagship cancer center. Trustworthy means the tool doesn't quietly fabricate the comfort we never offered. Patient-centered means the technology bends toward the human, not the other way around.
We cannot rewrite every line of code, but we can shape what we ask of it.
A grounded next step
Before your next session — whether you're a patient, a caregiver, or a clinician — try this. Name one question you want any AI-assisted tool to answer for you, out loud. What happens to what I share? Who reviews this? Can I opt out, today, without losing continuity of care? Saying it aloud turns an abstract concern into a concrete right. That single question, voiced clearly, is where safe care begins to anchor.