What "fact-check mode" actually does
It's worth being precise about what this feature is and isn't, because "fact checker" tools online range from careful source-weighing assistants to badge generators that output a confident-looking true/false with no real evidence behind it. @vustSearchBot's /fact mode is the former: send it a claim — a sentence someone forwarded you, a headline, something you half-remember from a documentary — and it runs a live web search specifically framed around evaluating that claim, then returns one of four verdict bands plus a short rationale and the sources it used. It does not consult a static database of "already checked" facts, and it does not pretend to be a court of final record. It reads current web sources and tells you, honestly, what they say right now.
That last phrase — "right now" — matters more than it might first seem. A claim about a scientific consensus, a historical event with settled documentation, or a well-established statistic will usually get a clean Supported or Unsupported verdict, because the source landscape for those topics is stable. A claim about a fast-moving news story, an early-stage scientific finding, or something genuinely contested between credible outlets will more often land on Mixed/Contested or Unverifiable — and that's the tool working correctly, not failing. A fact-checker that always gives you a confident yes or no is hiding the cases where the honest answer is "it's complicated" or "we don't know yet."
The four verdict bands, explained properly
Each check returns exactly one of four bands, chosen by the model after it has read the search results, not before:
- Supported — credible, independent sources confirm the claim. This is the band you'll see for well-documented facts: established scientific findings, verified historical events, statistics from primary sources like government agencies or peer-reviewed studies.
- Mixed / Contested — credible sources disagree with each other, or the claim is only partially true. This shows up often for claims that conflate a kernel of truth with an exaggeration, or for genuinely disputed topics where reasonable experts land on different sides.
- Unsupported — credible sources contradict the claim. This is the band for popular myths, outdated information that's since been corrected, and claims that sound plausible but don't hold up against documented evidence.
- Unverifiable — there isn't enough source coverage to judge the claim either way. This happens for very recent events that haven't been covered yet, extremely niche claims with no public documentation, or predictions about the future that can't be fact-checked by definition.
The fourth band is the one most "instant fact-checker" tools skip, because it's less satisfying than a clean verdict. But forcing a Supported/Unsupported answer onto a claim with genuinely insufficient evidence isn't more useful — it's less honest. If the sources aren't there, the tool says so instead of guessing.
How it fits into the rest of @vustSearchBot
Fact-check mode isn't a separate product bolted onto the bot — it deliberately reuses the exact same Quick-tier search pipeline that powers a normal /search query. Same 1✦ price, same daily Free Unlock allowance (five free Quick or Standard checks a day once you've joined the channel), same anti-abuse guard that filters obvious spam or injection attempts before anything reaches the search provider. If you're already a @vustSearchBot user, there's nothing new to learn about pricing or limits — /fact is just a different lens on the same underlying search.
This matters because it means fact-check mode inherits the same reliability guarantees as the rest of the bot: your balance is only ever charged after the check completes successfully, a failed or malformed provider response never silently burns your spark, and the same rate limiting that protects the bot from abuse protects fact-check requests too. There's no separate "fact-check tier" with its own pricing quirks to learn.
A worked example, start to finish
You send:
/fact drinking 8 glasses of water a day is a scientifically proven requirement
You get back:
⚖️ Mixed / contested
The "8 glasses a day" figure isn't a proven universal requirement — hydration needs vary by body size, climate, and diet, and a meaningful share of daily water intake comes from food, not just drinking water. Health bodies generally treat the figure as a rough, easy-to-remember guideline rather than a scientifically mandated minimum.
Based on a snapshot of current sources — not a final ruling. Re-check if the topic evolves.
Tap Sources under the result and you'll see the actual articles and health-body guidance the verdict drew from — the same source display used across every @vustSearchBot result, not a fact-check-specific feature. That consistency is deliberate: you shouldn't have to learn a new interface just because you're checking a claim instead of asking a question.
Why the honesty note matters
Every fact-check result ends with a fixed line: a reminder that the verdict is based on a snapshot of current sources, not a final ruling, and that it's worth re-checking if the topic evolves. This isn't legal boilerplate for its own sake — it's a genuine description of how the tool works. A claim checked today about an emerging scientific finding could get a different verdict in six months once more research is published. A claim about an ongoing news story could shift from Unverifiable to Supported or Unsupported once more reporting comes out. The tool is honest about that limitation instead of presenting its output as timeless truth.
How this differs from an "instant AI fact-checker" badge
A lot of consumer fact-checking tools optimize for a satisfying, shareable output: a green checkmark or red X, sometimes with a percentage confidence score attached that implies more precision than the underlying evidence supports. @vustSearchBot's approach deliberately resists that pull in three ways. First, the four-band system includes an honest "we can't tell" option instead of forcing every claim into a binary. Second, sources are always one tap away — you're never asked to trust a badge without being able to check the reasoning yourself. Third, there's no numeric confidence score dressed up as false precision; the rationale is qualitative and grounded in what the sources actually say, not a manufactured percentage.
What it won't do
Fact-check mode won't verify claims about you personally, private information, or anything that isn't discoverable through public web sources — it's a web-search-backed tool, not an investigator with access to private records. It also won't adjudicate genuinely subjective claims ("this is the best pizza in the city") since those aren't factual claims with a verifiable answer in the first place; feeding it an opinion phrased as fact will usually surface that ambiguity in the rationale rather than force a verdict. And it isn't a substitute for domain-expert review on high-stakes claims — medical, legal, or financial decisions deserve a professional's judgment, with a fact-check as one input among several, not the final word.
Getting the most useful verdict
A few habits make the check more useful. State the claim as a specific, checkable assertion rather than a vague impression — "this brand of supplement cures anxiety" gets a sharper check than "supplements are good for you." Include enough context to disambiguate what you mean if the claim could refer to multiple things — a date, a place, or a specific person's name where relevant. And if the verdict comes back Unverifiable, that's often useful information on its own: it tells you the claim is either too new, too niche, or too poorly documented for anyone (including you) to responsibly treat as settled fact yet.
Common claim patterns and what to expect from each
Not every claim behaves the same way once it's put through a source check, and knowing the rough pattern in advance helps you interpret the verdict correctly rather than being surprised by it.
Long-settled facts and statistics. Claims about well-documented history, established science, or figures from primary sources (a national statistics office, a peer-reviewed study, an official government record) tend to resolve cleanly to Supported or Unsupported, because the source landscape is stable and rarely contested by credible outlets. If you ask about something like the boiling point of water at sea level or the outcome of a well-documented historical event, expect a confident, quick verdict.
Popular myths and outdated corrections. A surprising number of claims that "everyone knows" turn out to be outdated or simply wrong once checked against current documentation — the kind of thing that spreads through repetition rather than evidence. These usually land as Unsupported, often with the rationale explaining exactly where the misconception came from and what the actual documented reality is.
Contested or politically charged claims. Topics where credible experts and institutions genuinely disagree, or where a claim conflates a true detail with a false generalization, tend to land on Mixed/Contested. This is the band to expect for anything where reasonable, well-sourced people land on different sides — the verdict won't try to force a side, it will describe the disagreement.
Breaking news and very recent events. A claim about something that happened in the last few hours or days often resolves to Unverifiable simply because credible reporting hasn't caught up yet, not because the claim is inherently unknowable. Re-running the same check a day or two later, once more coverage exists, can produce a different, more confident verdict — which is exactly why the honesty note reminds you the result is a snapshot in time.
Predictions and future-tense claims. Claims phrased as predictions ("this technology will replace X within five years") aren't factual claims in the ordinary sense — there's no current evidence that can confirm or deny a future event. These typically come back Unverifiable as well, with the rationale noting that the claim describes a future outcome rather than a checkable current fact.
Where fact-check mode sits relative to Search's other tiers
@vustSearchBot has four depth tiers for ordinary questions — Quick, Standard, Research, and the Pro-only Deep Report — and fact-check mode sits alongside them as a specialized lens on the Quick tier rather than a fifth tier. If a claim needs deeper investigation than a single search pass can offer — say, a genuinely complex historical dispute with dozens of competing sources — Research mode's multi-query pipeline is the better tool, since it plans multiple sub-searches and synthesizes across them instead of running a single pass. Fact-check mode is built for the common case: a specific, bounded claim that benefits from a structured verdict rather than an open-ended answer.