Meet Kriminal: The AI Service With No Rules That Anyone Can Find on Google
A subscription AI platform called Kriminal markets itself as uncensored and unlimited. Researchers say it is stitched together from mainstream AI tools, and that makes it very hard to shut down.
Key points
- A platform called Kriminal AI sells monthly subscriptions starting at $12.99, with no content filters and no identity checks on payment.
- Security researchers at ThreatDown, Malwarebytes' business-focused research unit, published findings this week describing it as "one of the newest and most popular tools in the criminal AI market."
- Kriminal claims 2,300 active users and more than 18,400 messages processed to date.
- Despite its branding, the site sits on the open internet and turns up in a basic Google search.
- The platform appears to be built almost entirely from well-known, off-the-shelf AI services rather than any proprietary technology.
Kriminal is easy to find. Type the name into a search engine and the website comes up. That alone sets it apart from the typical underground criminal marketplace, which usually hides on encrypted networks ordinary browsers cannot reach.
Yet the platform's pitch is anything but ordinary. It advertises features with names like WRAITH, described as a tool for "social engineering and persona craft," meaning it could help someone construct fake identities or deceptive scripts. Another feature, called ARCHITECT, is described as an "offensive security and exploit expert," which, in plain terms, means it claims to help users find and use weaknesses in computer systems. The platform also advertises the ability to generate images without filters, scan publicly available information for people's names and addresses, and trace cryptocurrency transactions.
Is this actually illegal?
Not necessarily, and that is exactly what makes it complicated. Kriminal's terms of service say the platform is for "research, creative, and educational purposes" and explicitly bans activities that break local, national, or international law.
OSINT, which stands for open-source intelligence, meaning the gathering of information from public sources, is used every day by journalists, private investigators, and security professionals. Offensive security testing is a real and legitimate career. The features Kriminal advertises are not, by themselves, evidence of a crime.
But the platform takes no steps to verify who is using those features or why. Payment is accepted only in cryptocurrency, with no KYC, meaning no "know your customer" identity checks that regulated financial services are legally required to perform. The operators do not publicly identify themselves anywhere on the site.
Why is it so hard to take down?
Because it is not one thing. According to ThreatDown's analysis, Kriminal appears to be assembled from a collection of mainstream services.
| Component | Apparent role |
|---|---|
| Grok | Primary AI model |
| Anthropic Claude | Long-context processing |
| Llama via OpenRouter | Specialist tasks |
| Google Cloud / Cloudflare | Hosting and traffic |
| Tavily | Live web search |
| NowPayments | Crypto checkout, no ID checks |
Each of those vendors sees only its own small piece of the operation. Cloudflare handles traffic but cannot read what the traffic is for. The payment processor records a cryptocurrency payment but has no view of what was purchased. As ThreatDown put it, "the takedown surface isn't a bulletproof host to seize: it's a dozen separate abuse-desk tickets, each addressing a fragment of an operation none of them can see in full."
That structure raises real questions under the terms of service each AI provider publishes, but proving a violation would require examining specific API usage patterns and outputs. No single regulator or company currently holds the full picture.
For ordinary people, the practical concern is less about the platform itself and more about what someone could do with it. If a person receives an unexpectedly persuasive message asking for personal information, or notices that someone seems to know their address without explanation, tools like this are part of the reason why.
Common questions
Could the AI companies whose tools are being used here stop this?
They could try, but each provider only sees its own layer of activity. Coordinated action across multiple vendors would be needed, and no enforcement mechanism currently requires that.
What should ordinary people do?
Be sceptical of unsolicited messages that feel unusually personal or persuasive, even if they appear to come from a known contact. If someone asks for your address, bank details, or login credentials out of context, verify through a separate channel before responding.
