SEON is expanding across multiple dimensions of identity risk, leaving AI-generated fraud nowhere to hide.
Creating a fake identity from scratch used to take a lot of work. A fraudster had to register the accounts, build up the history, complete each step of the workflow personally and do it slowly enough to avoid tripping a velocity rule. That labor was the constraint. It also created mistakes, such as a device that doesn’t match the location associated with the account or a journey that moved too quickly to be human. This is the basic inconsistency that most fraud stacks were built to catch.
That labor constraint is disappearing.
Now all it takes to fake a real customer is an internet connection, an AI tool and a simple prompt to fabricate a synthetic identity from nothing or hijack a real one. AI technology also enables the swift production of convincing fake IDs and deepfakes. While the fraudster still decides on the identity and target, an AI agent can register the account, fill out forms and navigate checkouts on their behalf at machine speed. Now faster and more consistently than a person ever could.
TL;DR:
Trusting the human behind an account has never been harder.
The Financial Action Task Force’s (FATF) Horizon Scan on AI and Deepfakes found that a smartphone is now enough to produce a convincing deepfake in minutes, as criminals combine real and fabricated identity data to create synthetic identities capable of opening accounts and evading detection controls. ACAMS’s Global AFC Threats Report 2026, surveying anti-financial-crime professionals across more than 200 jurisdictions, found 75% rate the malicious use of generative AI as high or very high risk, the top external threat to their programs for the third year running.
AI can make each signal look real. The full pattern is harder to fake.
One signal is not an identity story
Individual evidence can pass in isolation, with emails, phone numbers, addresses, devices and documents each appearing legitimate on their own. A fraud program that evaluates one clue at a time will clear all bad actors through onboarding. The one thing about fraud is that it leaves seams. Synthetic identities create inconsistencies across dimensions and time, and fraud rings reuse infrastructure across accounts built to look unrelated. Point-in-time checks, refreshed on a lag by a shared data aggregator and evaluated one dimension at a time, aren’t built to catch the various signals. The contradictions and relationships between clues remain hidden precisely because nothing is checking them against each other.
The fraud ring hasn’t changed what it needs to hide. It still needs the same address, device and account history to reappear across identities that are supposed to look unrelated. What’s changed is how fast and cleanly an agent can now paper over that — reusing an identity or account without a trace — unless the signals watching for it run deep and connected enough to catch it.
Signal depth and breadth across multiple dimensions closes the gap
A fraudster directing an AI agent can fake individual signals, but faking every dimension at once is a different problem — and that’s the problem SEON’s signal expansion is built to defeat.
SEON’s signal foundation has expanded from 900+ to 1,100+, across digital footprint, phone and carrier data, address, device and session behavior. Broader access across these dimensions creates a fuller identity graph that can surface fraud patterns faster and expose inconsistencies in a person’s identity and shared fraud infrastructure, without adding a new step to the customer journey.
Two things make that possible: signal depth and signal breadth.
- Signal depth is the level of context and specificity available within one dimension. It tells an analyst more about what a particular observation means, such as understanding information associated with an email, including how many data breaches it has been associated with, email age and email string analysis will tell you a lot more than whether an email address is deliverable. Then adding on online profiles associated with that email tells more insight into a person’s history and broader digital footprint.
- Signal breadth is the range of distinct, independently generated dimensions available across the customer journey. It allows a risk analyst to compare multiple dimensions together in context to assess if the information about a user makes sense. Different dimensions include email, phone, online presence, addresses, payments, network, location, behavior, device characteristics and integrity, to name a few, that can all credibly belong to the same identity.
Depth catches a weak signal. Breadth catches the contradictions between them and that’s what surfaces the identity that doesn’t add up.
More signals doesn’t mean greater friction
Additional signals across dimensions do not change the customer’s experience. SEON still collects data at the same points it always has — at signup, login and at the point of transaction or checkout. What expands is what SEON sees and connects behind each step: a real-time read on customer intent.
Better ways to tell a real customer from a fake one
SEON’s digital footprint coverage expanded to cover 350+ of the platform’s 1,100+ signals, spanning across a broad range of online profiles, such as entertainment, technology, dating, eCommerce, social media and travel.
New checks for email addresses include AI developer platforms (DeepSeek, Lovable, Windsurf), and broader signals include social and dating apps (Meta, OkCupid), U.S. real estate sites (Realtor, Redfin) and job boards (Glassdoor, Indeed).
New online profiles associated with a phone number checks messaging and social apps (GroupMe, MeWe, Foursquare), betting and gaming platforms (SportyBet, Strendus) and everyday business tools (Expensify, Zoho, Office365).
Online profiles are hard to fake because they take time to accumulate. A real person racks them up just by going about their life, using dozens of unrelated services and leaving a digital footprint they never set out to build. A fabricated identity has to build that identity deliberately from scratch for every account in the ring, which is where the effort stops being worth it. And that’s why deepfakes and synthetic identities lack a digital past.
Additional signals related to a customer’s phone, such as SIM swap and porting detection, help protect against ATOs by flagging a number the moment it’s swapped or transferred, not days later and also provide insight into the phone number’s carrier history.
Turn physical locations into a risk signal
Addresses are a great example of a data point that is collected but rarely analyzed. Turning it into an active risk signal starts with canonicalization, so every formatting variant of the same address, Apt 1, Unit 1, #1, maps to one shared ID, regardless of abbreviation, casing or punctuation. That consistency is what makes the data actionable.
When a fraud ring reuses an address across accounts built to appear unrelated, cycling unit numbers to dodge detection, that reuse becomes visible rather than hidden by formatting differences.
Catch fraud mid-session
The same principle applies to what happens inside a session, not just to the history behind it. Full behavioral context across the customer journey provides a new layer of continuous trust to catch coached fraud, remote access and bot activity mid-session, in the moments that matter most, from onboarding and login to account recovery, checkout and payment. Automation, off-screen behavior and active calls surface during the session, so rules can automatically block suspicious activity.
What actually changed
Bad actors haven’t changed, but how they commit fraud has. Now it’s possible to direct a machine that moves faster and more consistently than any human, without the fatigue or hesitation that used to trip velocity rules.
Signal depth across multiple dimensions is built to close the gap by checking enough independent, first-party dimensions together at once, rendering it impossible to fake them at scale.
See what this looks like in practice in the Hidden Risk Files, where SEON’s fraud consultants walk through the uniquely suspicious signals that led to the takedown of a fraud ring.
