Check it before you post it on LinkedIn.
AI makes things up. CiteJury catches it.
Paste the claim or the AI answer. CiteJury finds real sources, seals them into a pack, and has three AIs read the same pack and answer on their own. Every line of the verdict links to its source.
Free to start: 50 credits with a work email, 30 with a personal one. A standard check is about 5 credits. If a check fails, you don't pay.
Recorded checks, replayed as they ran. Read the full write-up →
- 1Paste a claimA sentence, an AI answer, or a whole file.
- 2Sources are sealed, then 3 AIs read themClaude, ChatGPT and Grok read the same pack. Nothing can be added later.
- 3See the verdictTrue or not, why, a safe way to say it, and links to the proof.
Three real checks.
Open any of them.
Each verdict below came from a recorded run: real sources, real votes, nothing staged. Run one yourself to read every source and every juror's answer.
“Deloitte refunded the full A$440,000 fee to the Australian government over its AI-error report.”
The story is real, but the refund was partial, not the full fee.
Deloitte agreed to a partial refund, the final instalment of its A$440,000 contract, not a repayment of the full fee.
“Humans only use 10% of their brains.”
It's a myth. Neuroscience sources say the whole brain is used.
Imaging shows the whole brain is active, though not every region is busy at the same moment.
“will robots will take over humans in future”
Not settled; about 47 per cent of US employment at risk.
Say the pack does not show that robots will take over humans; the only clear share is one study summary that about 47 per cent of total US employment is at risk, and that is not a prediction those jobs will vanish.
Why not just ask one AI?
It can sound sure and still be wrong.
CiteJury finds real sources first. Then three AIs read them and answer on their own. You see where they agree.
Real mistakes
CiteJury caught.
The hardest mistakes to spot are the ones that are almost right. Each one here came from a real check.
“refunded the full A$440,000 fee”
→a partial refund, the final instalment
“settled with the EPA for around $3M”
→a $900,000 penalty plus up to $23M in fixes
“oil spill negligence that led to a fire”
→a fire ruptured hoses and spilled oil
“Blue Harvest acquired High Liner Foods”
→it bought High Liner's scallop business
Four verdicts.
No guessing.
Claude, ChatGPT and Grok each vote on their own. A final review checks that every quote they used is really in its source. Only then do you see the result.
See every step on a real check →Who uses CiteJury?
01Newsrooms
Journalists check a fact before the story goes out.
Paste the line. In about three minutes, three AIs tell you if it's right, and give you a safe way to say it.
How newsrooms use it →“Deloitte refunded the full A$440,000 fee to the Australian government over its AI-error report.”
The story is real, but the refund was partial, not the full fee.
Deloitte agreed to a partial refund, the final instalment of its A$440,000 contract, not a repayment of the full fee.
02Decks and reports
Analysts check every claim in a slide deck.
Upload a PowerPoint and pick up to 25 claims. Each one comes back marked, with a fix you can paste.
See the full report on this deck →“Trident settled with the EPA for around $3M under the Clean Air Act.”
Wrong amount: a $900,000 penalty plus up to $23M in fixes, not about $3M.
Trident settled Clean Air Act claims with the EPA, paying a $900,000 penalty and agreeing to spend up to $23 million on refrigerant fixes.
03Supplier risk
Buying teams check what a supplier says about itself.
Deep checks search wider and add Gemini as a fourth AI. Good for niche, legal or very new topics.
Supplier checks →“OBI Seafoods was formed by merging Ocean Beauty Seafoods and Icicle Seafoods' Alaska operations.”
OBI was formed in 2020 from the two companies' Alaska salmon and groundfish units.
04Essays and research
Students catch myths before they go in the essay.
A standard check costs about 50¢ and takes a minute or two. Every line of the answer links to a source.
For students and researchers →“Humans only use 10% of their brains.”
It's a myth. Neuroscience sources say the whole brain is used.
Imaging shows the whole brain is active, though not every region is busy at the same moment.
Pay for what a check uses.
Buy credits in packs. They never expire. A standard check costs about 5 credits ($0.50). Work emails start with 50 free credits, personal ones with 30.
Pro
$50 to $2,500 + tax where applicable
500 to 25,000 credits
Choose the amount you need. Same price per credit.
Get startedPrices are in USD. In the EU, the UK, Norway, Iceland, Switzerland, Australia and New Zealand they include VAT or GST; elsewhere any sales tax is added at checkout.
Typical cost
| Reading a documentThen each claim you pick is one check | about 1 credit |
| Quick1 AI, no final review | about 2 credits |
| StandardClaude, ChatGPT and Grok, plus a final review. Claude and ChatGPT also search the web | about 5 credits |
| DeepWidest search. Gemini joins as a fourth AI, and Claude, ChatGPT and Grok also search the web | about 10 credits |
| Company searchStandard, reads many company pages | about 30 credits |
| Follow-up questionAnswered from the same sources | about 1 credit |
- You only pay what a check actually uses
- Failed checks are free
- Reopening and downloading are free
- Credits never expire
- We don't keep your files
Your files
stay private.
We read your file, then delete it right away.
The text from your file is deleted once your claims are checked.
We don't use your content to train AI.
How we handle your data →Good
questions.
How is this different from asking ChatGPT?+
ChatGPT is one AI giving one answer. CiteJury first finds real, dated sources. Then it gives the same sources to three AIs (four on Deep) and checks every quote they use. You see where they agree, where they don't, and why.
What if the sources don't settle it?+
You get Can't confirm, not a guess. If good sources say different things, you get Sources disagree and see both sides.
What happens to documents I upload?+
We read the file and delete it right away. The text from it is deleted once your claims are checked. We keep only each claim, the words it came from, its result and its sources.
Which AIs do you use?+
Claude, ChatGPT and Grok each read the same sources and answer on their own. A final review then writes the result. Deep checks add Gemini as a fourth AI. A small helper model called Jev (from TypeSafe) removes off-topic sources first. It never decides the result.
Can CiteJury be wrong?+
Yes. A result is only as good as its sources, so every line links to one. Open them before you rely on it. If something looks wrong, use Report a problem on the result page.
Do I get free credits?+
Yes. Sign up with a work email and you get 50 free credits, enough for about 9 checks. Personal emails like Gmail or Outlook get 30. No card needed.
Check the facts before you post, publish or hit send.
Free to start: 50 credits with a work email, 30 with a personal one. If a check fails, you don't pay.