Faking a customer’s identity used to take real work. Now it takes little more than an internet connection and a well-crafted prompt. That shift is exactly what sits behind SEON’s latest move: the fraud prevention platform has expanded its SEON AI fraud detection signal library from more than 900 data points to over 1,100, adding new layers of address, device, phone and session intelligence designed to catch identities that look real but were built in minutes rather than lived over years.
Key takeaways
- SEON grew its proprietary signal foundation from 900+ to more than 1,100 directly sourced data points.
- New signals cover address intelligence, session behavior, phone and carrier data, digital footprint and device signals.
- Address Intelligence now verifies and standardizes addresses across 240+ countries.
- ACAMS says 75% of anti-financial crime professionals rank GenAI misuse as their top emerging risk for a third straight year.
- The new signals feed into SEON’s AI Command Center, where they support rules, alerts, reviews and investigations.
SEON expands proprietary fraud detection signals
SEON, which brands itself as an AI Command Center for fraud prevention and AML compliance, says the expansion is meant to help risk teams catch identity inconsistencies and the shared infrastructure that links fraud rings together, without adding friction to the customer journey. The company frames this as a direct response to how cheap and fast it has become to manufacture a convincing digital persona.
From 900+ to 1,100 signals across address, device, phone and session data
The expansion spans five broad categories: address intelligence, session behavior, phone and carrier data, digital footprint checks and device signals. Each layer is designed to be checked against the others, since SEON notes that any single signal — a clean email, a legitimate-looking device, a verified address — can pass inspection on its own while still being part of a fabricated identity once cross-referenced with the rest.
Why AI-generated fraud is outpacing traditional detection
Generative AI has collapsed the time and effort it takes to build a fake identity that looks legitimate on paper. Where fraudsters once had to assemble documents, fill out forms and operate accounts one by one, GenAI tools can now produce believable profiles and consistent device histories in a matter of minutes. That speed is precisely what worries regulators and compliance professionals watching the space.
The Financial Action Task Force has warned that anyone with a smartphone can generate a convincing deepfake in roughly the time it takes to set up a social media profile. Industry group ACAMS backs that concern with data: 75% of anti-financial crime professionals now rank GenAI misuse as their top emerging risk, a position it has held for three consecutive years. Fraudsters still choose the identity and the target — but increasingly, AI agents handle the production work at machine speed.
Enhanced fraud detection through detailed signal layers
SEON’s expanded signal set is built around three questions investigators keep coming back to: does this identity have a real history, is it connected to infrastructure used elsewhere, and how is it behaving right now. Each new signal category targets one of those questions directly.
Verifying history through digital footprint and phone data
Digital Footprint checks now trace where an email address or phone number has shown up across services over time, with expanded coverage that includes AI developer platforms, job boards, real estate sites and dating apps. Alongside that, Phone Intelligence adds SIM-swap and porting history, giving risk teams a way to judge whether a phone number and its associated identity have actually existed before this particular transaction — rather than appearing out of nowhere.
Exposing shared infrastructure via address and device intelligence
Address Intelligence now verifies and standardizes addresses across more than 240 countries, assigning consistent identifiers to a specific address and building rather than treating every input as a fresh, unrelated string of text. That makes it possible to spot when supposedly unrelated accounts keep cycling through the same location using different unit numbers or formatting tricks. Device Intelligence, meanwhile, has been expanded to surface AI-agent activity, compromised iOS devices, mismatches between Android eSIMs and network-country discrepancies that appear when a VPN is masking the real IP address.
Tracking live behavior with session monitoring
Session Monitoring follows a customer’s behavior from onboarding through login, account recovery, checkout and payment. This lets fraud teams detect automation, remote access, off-screen activity or an active call happening in the background while a session is still live — giving them a chance to intervene before it escalates into a full account takeover.
Integrating signals into SEON’s AI Command Center
None of these signals work in isolation. Inside SEON’s AI Command Center, fraud and risk teams can plug the new data points straight into rules, alerts, customer reviews and network investigations, so every decision — whether made by a human analyst or an automated system — draws on the same evidence base. SEON has also made every signal in the expansion available through its Model Context Protocol server, letting investigators connect the data to their own AI tools so those systems can reason from real evidence rather than incomplete snapshots.
This matters for anyone trying to understand where fraud prevention is heading: as GenAI lowers the cost of creating a fake identity, the competitive edge shifts toward whoever can verify history and cross-reference infrastructure fastest. SEON’s bet is that stacking more independent, directly sourced signals against each other — rather than relying on any single check — is what makes reused infrastructure and manufactured histories harder to hide.
What SEON’s CEO says about faking believable histories
Tamas Kadar, CEO and Co-Founder of SEON, put the core problem in blunt terms. “AI has made a believable identity cheap to produce. What fraudsters cannot easily do at scale is build a consistent history for every account without reusing infrastructure. That is where our signal foundation makes the difference. The more dimensions a fraud team can check simultaneously, the harder it is to hide an identity that does not add up,” Kadar said.
SEON has paired the signal expansion with a new editorial project called Hidden Risk Files, a series of short investigations from its fraud consultants showing how specific signals catch fraud that other checks miss. The first installment reportedly traces how a single device attribute — a screen-brightness reading — connected thousands of accounts inside a fraud ring running on real Android hardware.
FAQ
What new data points has SEON added to its fraud detection platform?
SEON expanded its signals to include address intelligence, session behavior, phone and carrier data, digital footprint, and device signals.
How do AI-generated fake identities challenge fraud detection?
GenAI tools can create believable profiles and consistent device histories in minutes, making it harder to detect fake identities.
What does the Financial Action Task Force say about AI-generated deepfakes?
According to FATF, virtually anyone owning a smartphone is able to produce a convincing deepfake in roughly the same amount of time needed to create a social media account.
How does SEON’s signal intelligence help expose fraud rings?
It exposes identity inconsistencies and reused infrastructure linking fraud rings by verifying multiple data layers like addresses, devices, and phone histories.
Article produced with the assistance of artificial intelligence and reviewed by the editorial team.