Question 1

How is this different from a content moderation API?

Accepted Answer

A moderation API reads the text and returns a policy category. That catches profanity and gore, and misses the scam entirely, because a sophisticated scam contains no banned words. It is a plausible listing with an implausible way to pay, or a login page that is perfect apart from the domain. Rupt reads for intent rather than policy, then adds what no text model can know: how many identities that browser has carried, whether the email domain was registered last week, and how many other accounts the same person is running right now.

Question 2

Can it catch phishing pages our users deploy on our product?

Accepted Answer

Yes, and this is one of the main reasons app builders and hosting platforms run it. Rupt reads what your users ship, not just what they type, so a cloned bank login, a wallet drainer behind a launch page, or a storefront built only to harvest card details is caught before it goes anywhere. The same account scoring applies, so the operator who deploys one phishing site and the twelve accounts they have queued up behind it come down together.

Question 3

What surfaces can it screen?

Accepted Answer

Anything your users create. Marketplace listings, job and task posts, service offers, direct messages and chat, comments, reviews and ratings, profile bios and display names, and the sites, apps and pages built on your product. You set the bar per surface, so a first review is treated differently from a first payment request.

Question 4

Does it work when a real human writes everything by hand?

Accepted Answer

That is the case it is built for. We have tracked actors who wrote every message themselves, at ordinary human speed, one new account a day for months. No bot check fires and no velocity rule fires. What gives them away is reuse: one browser carrying nineteen identities, one IP carrying eighteen signups, a discarded phone number tying six accounts together. Never read no bot signals as no fraud.

Question 5

Will it flag my legitimate users?

Accepted Answer

Rupt starts conservative and hands you the evidence rather than a bare verdict, so a borderline item goes to your queue with the reasons attached instead of straight to a takedown. Because the score combines content with device, email and network intelligence, a real user writing an unusual message does not look like a scammer, and you are not tuning a text threshold on its own.

Question 6

Do we still need human moderators?

Accepted Answer

Most teams keep people in the loop and use Rupt to decide what reaches them. The obvious scams are held automatically, the clean traffic never enters the queue, and what is left arrives with the linked accounts, the device history and the network evidence already attached. The work moves from reading everything to deciding on the few that matter.

Question 7

Does it slow down my product?

Accepted Answer

No. Screening returns in under a second and runs alongside the request rather than in front of it, so you can hold an item on the verdict or publish and act after. The account scoring you already have from signup, so there is nothing extra to compute at post time.

Question 8

How long does it take to integrate?

Accepted Answer

Most teams ship in a day. Drop in the SDK for web, iOS or Android, send the content you want read, and act on the verdict. If you already run Rupt for fake accounts or account sharing, the account intelligence is there and this is a policy change.

Question 9

What does it cost?

Accepted Answer

Premium starts at $99 a month, which includes credits you spend across evaluations, screening and executed challenges. Screening scales with the content you send, and most platforms only escalate a small share of it.

Detect

Act

Free tools

By use case

By industry

Developers

Learn

Stop scams, spam and phishing on your platform.

Whatever your users create, a marketplace listing, a job post, a message, or an entire site built on your product, Rupt reads it for fraud and scores the account behind it.

Fraud does not only arrive as a message.

Marketplaces and task platforms

Products your users build on

User to user communication

Platforms that stopped treating every account as trustworthy:

Built for the scams that read as legitimate.

An AI that reads for intent, not for keywords.

The post is the symptom. The operator is the problem.

You decide what a verdict is worth.

The scam that no content filter was ever going to catch.

How it ran

Signed up

Bought a burner

Rotated networks

Spammed the providers

Started over tomorrow

One signal is an opinion. Four is a case.

Content intelligence

Device intelligence

Email intelligence

IP intelligence

Questions teams ask before turning on scam and spam prevention:

Find out what is already running on your platform.