According to a recent report from DuckDuckGoose, AI-generated synthetic identities are circumventing digital onboarding systems in significant numbers across Latin America’s financial sector.

The research reveals that identity fraud has progressed beyond the manipulation of existing documents or images, now encompassing real-time creation of fully synthetic human identities that can pass standard onboarding and authentication controls.

There has been a swift rise in tools for producing synthetic media. In Q4 2025, more than 55 synthetic media generators were publicly released, roughly one every 1.6 days. Since the beginning of 2024, image-to-video generation capabilities have increased by over 1,000%, significantly impacting biometric checks and liveness detection—two crucial controls in digital financial onboarding.

Open AI ecosystem monitoring uncovered nearly 868,000 unique synthetic identity variants being created monthly. Each variant introduces characteristics that verification systems are yet to encounter, creating opportunities for fraudulent identities to slip through controls designed against earlier fraud patterns.

This is presented as a structural challenge rather than a temporary issue. As the velocity of generator releases continues to rise, detection models face an ongoing gap in their capabilities, and verification environments must adapt to a continuous influx of novel synthetic profiles instead of relying on static sets of known attack strategies.

Focus on Brazil and Latin America

The report highlights specific risks associated with Brazil and the broader Latin American market. The rapid deployment of PIX, Brazil’s real-time payments system, along with widespread remote onboarding and the growth of neobanks, has positioned identity verification at the core of financial security frameworks.

In 2024, Brazilian institutions reported USD 1.9 billion (BRL 10.1 billion) in banking fraud losses, with fraud execution times now measured in minutes rather than days. These conditions exacerbate the operational risks posed by any lag between synthetic identity generation capabilities and detection infrastructure.

Implications for Identity Infrastructure

The report concludes that traditional Know Your Customer (KYC) and liveness verification measures, while effective as designed, are not equipped to handle identities continuously generated and adapted at a rate faster than detection models can be updated. This changes the underlying assumptions of these controls in an environment where AI-generated synthetic presence is becoming increasingly real-time.

A company official quoted in the report emphasized that trust must be established during identity creation, rather than after fraud detection. For identity platforms operating within digital economies like Brazil, the report suggests that synthetic media detection has transitioned from an advanced feature to a necessary infrastructure component, on par with document verification and liveness checks as foundational elements of the onboarding process.