The explosion of generative AI has brought an unprecedented rise in deepfakes, impersonations, and automated fraud. In this scenario, Spanish startup Didit, founded by brothers Alberto and Alejandro Rosas, has closed a €6.4 million ($7.5 million) seed round to build a programmable identity infrastructure that promises to change the game. Backed by Y Combinator and with a new office in San Francisco, the company aims to become the global standard for verifying people, companies, AI agents, wallets, and transactions.

A capital injection to grow without losing profitability

Alejandro Rosas, co-founder and CEO, explained that the funds will be used entirely to hire talent in key areas such as sales, partnerships, product development, and institutional relations. "All this without ceasing to be profitable at any time and always maintaining a good margin," he assured. The opening of the San Francisco office is not merely symbolic: "Being here positions you as an international company and gives you access to top-tier tech talent, and also allows you to move at a much faster pace than in Spain," he added.

For clients, the US presence provides a direct competitive advantage: being able to hold in-person demos with CEOs in minutes, accelerating sales and adoption cycles.

What is programmable identity?

Didit was born from an internal need. The Rosas brothers were looking for an identity verification provider to build a universal wallet, but found closed, inflexible solutions. "If AI explodes, fraud will too, and bots will multiply. We saw that our idea made a lot of sense," Alejandro recounts.

Didit's platform is based on a single API that allows integrating identity verification into any digital product. Unlike traditional systems, which are often fragmented and hard to scale, Didit offers a centralized approach that analyzes over 200 signals, including biometrics, deepfake detection, and behavioral analysis.

"We've tried to make it as simple as possible. A good product simplifies complexity, both at the API and design levels," Rosas highlights. This philosophy allows developers to integrate only the functionalities they need, without being overwhelmed by unnecessary options.

Proprietary AI models and threat detection

The heart of the platform is the artificial intelligence models developed by Didit's team since 2017. These models range from extracting data from identity documents of any country to facial recognition with high precision in databases of millions of people. "We constantly retrain and improve because new attacks, methodologies, and generative models emerge every day," Rosas explains.

Didit's competitive advantage lies in its open platform: by allowing anyone to try to attack the system, the company gains valuable insights into emerging fraud techniques. "We see all attacks before they reach our clients," he states.

Verification of people, transactions, agents, and wallets

One of Didit's most innovative aspects is its ability to verify not only people but also transactions, AI agents, and wallets. This responds to a growing market demand: "Large banks and entities have multiple fraud providers. What they want is to unify everything," Rosas notes.

The platform centralizes initial verification data with subsequent transaction data, creating an ecosystem of signals that improves fraud detection. "By having everything in one place, we can identify patterns that would otherwise go unnoticed," he adds.

The seal of the Spanish Government

Didit's technology has been verified by the Spanish Government as equivalent to or more secure than in-person verification. How did they achieve this? "We hacked several identity providers in Spain to understand their weaknesses and then designed a system with maximum security," Rosas confesses.

The process includes reading the chip of identity documents (DNI or passport), verifying the digital signature of the National Police, and real-time biometric comparison. "It's not enough to have the physical document; we also check that the person is alive, that it's not a photo, a mask, or a deepfake," he details.

Balancing privacy, regulation, and compliance

In a context of increasingly strict regulations (such as the European eIDAS or privacy laws in different countries), Didit has opted for a local deployment approach. "We offer the same service, but replicated within each country so that data does not leave its borders," Rosas explains. This allows compliance with local regulations without sacrificing functionality.

For IT professionals and compliance officers, this architecture is key, as it simplifies data sovereignty management and reduces legal risks.

AI as an accelerator of fraud… and opportunity

"For us, the more AI grows, the better," Rosas states emphatically. "More fraud means more need for verification, and we have an open platform, competitively priced, with a growing network effect."

This network effect is based on the premise that the more clients use Didit, the more information is generated about fraudulent actors, which in turn attracts new clients. "In the end, relatively few generate most of the fraud. We are creating a global signal network to detect them," he concludes.

Didit's bet comes at a time when identity verification has become a critical pillar for digital transformation. Companies across all sectors—from fintech to social networks—need robust solutions to operate with confidence in a world where AI can generate content indistinguishable from reality.

For CTOs and security officers, Didit's proposal represents a natural evolution: moving from isolated systems to a unified infrastructure that not only verifies identities but also proactively prevents fraud. In a market where speed of adaptation makes the difference, having a single API and a constantly improving AI model can be the advantage many companies need.

If you are interested in the impact of AI on security and digital identity, don't miss our analysis on how ethical hacking helps prevent fraud, or discover how sustainable interconnection is changing the tech landscape.

Original source: ComputerWorld. Analysis and adaptation by ForgeNEX.