Use cases

When to check
before you trust.

Concrete situations, grouped by detector. Each one leads to the page of the detector that answers it.

Use cases in depth

Workflows for teams who need to check content for AI: what can go wrong, which checks to run and how to read the result.

Teachers and schools

An AI detector for teachers is a first signal that tells you which essays deserve a closer look. It is not proof. Pair it with process evidence (drafts, version history and a short conversation with the student) and never sanction a student on a detector result alone.

Hiring and recruiting

To detect AI in job applications, focus on the two risks that matter: a candidate who cannot do the work they submitted, and a candidate who is not who they claim to be. Use detectors on take-home tests and interview recordings to decide what to ask next, then confirm with live questions and identity checks.

Journalists and fact-checkers

The reliable way to verify images and videos for journalists and fact-checkers: trace the content to its first source, check its provenance and metadata, confirm where and when it was made, and then use AI detection as one more signal. A detector result supports your verification; it never replaces it, and you should report it as an indication.

Fraud and KYC teams

To detect AI-generated documents and images for fraud, combine data checks (does the merchant, policy or account exist?) with provenance and AI-generation analysis of the file itself. AnyFact flags signs of AI generation in photos, documents, video and voice; it is not an identity verification provider and does not check documents against official registries.

Fake dating profiles

A fake profile picture checker tells you whether a dating photo shows signs of AI generation. Many fake profiles use stolen real photos instead, so pair it with a reverse image search and a live video call. Whatever the result, never send money to someone you have not met in person.

Voice clone scams

An AI voice scam uses a cloned voice of someone you trust (a child, a parent, your boss) to pressure you into sending money or sharing codes. The best defense does not depend on the audio: hang up and call back on a number you already know, and agree on a family code word. A voice detector can then check a recorded message for signs of cloning.

Trust and safety moderation

AI generated content moderation works best when you moderate by harm, not by whether something is AI. Label disclosed or provenance-marked synthetic media, remove deceptive and abusive content, and use detectors to support human reviewers on escalated cases, never as an automatic judge.

Marketplaces and e-commerce

To detect AI generated product images, start with the listing around the photo: seller history, price, and whether the same item looks the same from every angle. Then check the photo itself for provenance data and generation signals, and ask the seller for a proof photo. A detector result tells you where to look; a human decides whether the listing misleads buyers.

Publishers and SEO teams

An AI content checker for publishers tells an editor which drafts, guest posts or pages deserve a closer read. It does not judge quality and it does not predict rankings. Use it alongside fact-checking and a clear AI clause in your briefs, and talk to the writer before you reject anything.

Academic publishing

To detect AI in research papers, compare the manuscript with the journal’s disclosure policy, verify its references and figures, and use a detector to find passages or images that need questions. Many journals allow disclosed AI help with language, so the real targets are undisclosed generation and fabricated content. A detector result is a reason to ask the authors, never grounds for rejection on its own.

Lawyers and legal evidence

To detect deepfake evidence early, preserve the original file, establish who made it and how, check its metadata and provenance, then screen it with an AI detector to decide whether a forensic examiner is needed. A screening result is not expert evidence. AnyFact is not a forensic service and its reports are not court-certified; for litigation, retain a qualified forensic examiner.

Elections and misinformation

To detect election deepfakes, trace the clip to its original source, check it against the candidate’s or election authority’s official channels, look for other recordings of the same event, and then run AI detection as one more signal. Apply the same standard to every party and candidate. For anything about when, where or how to vote, the official election authority is the source to trust.

Artists and art contests

An AI art detector for contests flags entries that show signs of generation so the jury knows which ones to ask about. It cannot prove how a piece was made. Combine it with clear entry rules, content credentials and process evidence such as sketches, layered files or RAW photos, and never accuse an artist publicly on a score.

Insurance claims

To detect AI generated insurance claim photos, check the claim facts first, ask for the original files from the claimant’s phone, compare metadata and the photos with each other, and screen suspicious images for signs of generation. AnyFact flags AI generation in photos, documents and voice recordings; it does not reliably detect manual edits and it is not a claims-fraud platform.

Is it legit? Check a business before you pay

Shops, tickets, resale apps, job offers, trading platforms and suppliers: the warning signs, the official rules and what our Business Check verifies. More in the legit checks.

Online shopping

A fake online store usually fails on three things a real shop cannot fake cheaply: a domain that has existed for years, a company you can find in a register, and a way to pay that you can dispute. Check the address, the company named in the legal notice and the payment options before you enter a card number; our Business Check runs those checks for you and links every source.

Resale marketplaces

Resale apps such as Depop, Mercari, Poshmark, StockX and Whatnot are established platforms; the risk is the individual seller, and the protection only works if you pay inside the app. Each platform gives you a short window to report a problem after delivery, as short as 72 hours on Mercari. Know the window before you buy, never pay a seller outside the app, and check any outside site a seller sends you to.

Event tickets

The large resale marketplaces (StubHub, Vivid Seats, SeatGeek, TickPick) back purchases made on their platform with a guarantee of valid tickets or your money back. Ticket scams happen around them: sellers on social media who ask for a bank transfer or a payment app, and unknown ticket sites that copy the big ones. Buy on the platform’s own checkout, and check any unfamiliar ticket site before paying.

Get-paid apps

A legit get-paid app names the company that runs it in its terms, states its cash-out minimum and payout methods up front, and never asks you to pay to receive your earnings. Freecash and KashKick, for example, both publish their operator and a minimum withdrawal. The scams are elsewhere: fake prize notices, “task” jobs that ask for deposits, and copies of known apps.

Job offers

Fake job opportunities give themselves away when money flows the wrong way: you are asked to pay for training or equipment, to deposit a check and send part of it back, or to fund “tasks” to unlock your pay. Before you send documents or money, check that the company exists and matches the recruiter’s email and website, and call it on a number from its official site.

Investment platforms

Forex trading is not a scam in itself, but unauthorised trading platforms are the costliest fraud reported to the FBI and the FTC. Before you deposit, check that the firm is authorised by the regulator of your country and is not on a regulator warning list; “AI trading bots” that promise steady returns are a pattern US regulators have warned about. Our Business Check searches the warning lists of the FCA, the AMF, CySEC, ASIC and others for you.

AI voice and deepfake scams

An AI voice cloning scam or a deepfake video call imitates someone you trust (an executive, a public official, a supplier) to get a payment or access. The defence that works does not depend on the audio or the image: end the call and call back on a number you already have, and verify the company behind any new request. A recording can then go through our voice and video detectors as a second check.

Suppliers and partners

To verify a supplier, confirm that the company exists in an official register, that its name, address and identifier match its website and invoice, and that the bank account belongs to it. Fake company names borrowed from real businesses and hijacked supplier emails are the two traps; the FBI recorded $3.05 billion lost to business email compromise in 2025.

AI Photo Detector

3 cases