Colombia warns of AI identity fraud
Incode warned that AI agents are using fake documents and synthetic identities against banks and fintechs in Colombia.
Criminals are using artificial intelligence to generate fake documents and images, open accounts, and adapt their attacks when they are blocked, according to an alert circulated by Incode and picked up by media outlets in Colombia. The warning points to banks, fintechs and users, with device signals, attempt frequency and identity history as key detection indicators.
Criminals are using artificial intelligence to create fake documents and images, try to open accounts, and change tactics when they are rejected, according to an Incode alert circulated in Colombia. The warning puts banks, fintechs and users at risk, and points to detection controls based on device signals, attempt frequency and identity history.
What does Incode describe as agentic fraud?
Incode defines agentic fraud as fraud generated, assisted or operated by AI. The examples it cites include fake documents, cloned voices and synthetic faces, a set of techniques the company links to account-opening attempts and to attacks that adapt after each rejection.
The warning was repeated by FOLOU and by Full Magazine Noticias, which also noted that AI agents can adjust their behavior after each failed attempt. Along those lines, the version reproduced by Full Magazine Noticias describes controls such as device signals, attempt frequency and identity history to detect anomalous behavior at financial institutions in Colombia.
What is TrustGraph and how is it used?
Incode described TrustGraph as a network for connecting fraud and trust signals across participating institutions. According to the version reproduced by Full Magazine Noticias, it allows matching patterns without centralizing customer records or enabling one institution to query another institution's records.
Incode's technical documentation adds that the system looks for matches with identities, devices and documents flagged as fraudulent, along with unusual activity, registration speed and possible fraud networks. That approach is meant to cross scattered signals between institutions without moving sensitive databases from one actor to another.
What regional scope did the warning show?
FOLOU said Incode identified one Latin American case in which a single attacker was linked to 426 verification sessions, 35 documents and activity across eight fintechs. The example was presented as a regional case, but the coverage itself did not confirm that this specific case happened in Colombia.
At the same time, AsuntosLegales.co published Colombia-focused coverage that linked the growth of digital fraud and the use of artificial intelligence with digital identity protection. That report attributed 58,366 cases recorded through September 2026 to figures from Asobancaria and the Financial Superintendence, in a context where document fraud and impersonation are becoming bigger concerns for the sector.
What do the available materials say about detection?
Incode's documentation describes concrete controls for finding fraud through matches with identities, devices and documents flagged as fraudulent, as well as signs of unusual activity. The combination of those indicators is the basis the company proposes for detecting abuse patterns in verification sessions and repeated registrations.
In a corporate post, Incode said TrustGraph had reportedly helped detect up to 53% more fraud than existing models during early access. That result is not independently corroborated in the available materials, so it should be read as a company claim, not outside validation.
Sources
- Delincuentes usan inteligencia artificial para crear identidades falsas: Incode advierte sobre los riesgos para Colombiafullmagazinenoticias.blogspot.com· Full Magazine Noticias
- Con el fraude digital, ¿qué tan seguros están sus movimientos e información bancaria?asuntoslegales.com.co· AsuntosLegales.co
- What is agentic fraud? How it works and how to stop itincode.com· Incode
- Incode ranks in the Top 5 of TechRound's Cybersecurity40incode.com· Incode
- Alerta en Colombia: uso de IA para falsificar identidadesfolou.co· FOLOU
- View Session Detailsdeveloper.incode.com· Incode


