Skip to content
Check it yourself

AI Assistant Claim: How to Check What’s True

Every verdict carries a confidence score and the sources behind it.

About this check

When an AI assistant makes a claim, the useful question is not only “What did it say?” It is “What evidence supports that answer, and can I check it myself?”

The direct path is to submit the claim to VOM. You can enter it by voice, and VOM checks it using real-time web search. The response includes three things to look for: a verdict, a confidence score, and the sources used to reach that verdict. Those sources are available for you to open and review yourself.

The verdict is not limited to an automatic yes or no. VOM can report a claim as true, false, or inconclusive. When the available evidence cannot settle the claim, it is reported as inconclusive rather than being forced into one of the other two categories. That distinction matters for claims that are incomplete, disputed, or not supported by enough evidence.

The same approach applies when the claim arrives as an image. VOM can read the text inside a screenshot, so a forwarded image can be checked without retyping it. It can also assess an uploaded image as likely authentic, likely AI-generated, or unverifiable. These are separate questions: what the image says, and whether the image itself appears authentic.

The manual method

You can also check an AI assistant’s claim yourself. Start by preserving the exact wording. Small changes in wording can change what you are checking. Then search for the claim and open the sources behind the results rather than relying only on a headline, summary, or snippet.

Read enough of each source to identify what it actually supports. Compare the wording of the claim with the evidence. Check whether the source addresses the same subject, date, location, and scope. If sources disagree, record the disagreement instead of treating the first result as decisive. At the end, classify the claim as supported, contradicted, or unresolved.

That process is useful, but it has clear stopping points. A search result is not the same as evidence. A headline may leave out important context. Different sources may address different versions of a claim. A screenshot may contain text without showing where the text came from. And when the available material does not settle the question, a confident yes or no would go beyond the evidence.

VOM puts the checking step and the supporting material in the same response: a verdict, a confidence score, and cited sources. It can also hold a natural back-and-forth conversation for everyday questions, learning, brainstorming, writing, and planning. That means you can ask a follow-up question when a claim needs clarification instead of treating the first answer as the end of the process.

The wider workflow is not limited to checking text. VOM offers local, world, and trending feeds of circulating claims. It can turn a written idea into a generated image, and it independently reviews an AI-generated image or video before approving it for publication. Its article-sharing pipeline still publishes if AI image or video generation is blocked, fails, or is exhausted.

VOM answers in the language a person is using across its eight supported languages. The core task remains the same: separate the claim from the evidence, show the basis for the answer, and leave room for an inconclusive result when the evidence does not justify certainty.

Keep reading