Artificial intelligence-native infrastructure provider for identity and fraud prevention Didit today announced it has raised $6 million in seed funding.

New and continuing investors joining the round included Y Combinator LLC, Pioneer Fund, Orange Collective, Founders Future, Phosphor Capital, SaaSholic and Rebel Fund, alongside angel investors of Gusto Inc. Co-founder Tomer London, Fond Technologies Inc. Founder Taro Fukuyama among others. The round brings the total raised by the company to $7.5 million.

Didit’s vision is to build programmable identity infrastructure for the internet. This means a developer-first way to verify people, businesses and digital actions at a global scale when some entities aren’t even human – generative AI agents are changing the very landscape of what it means to take action on the web and within corporate networks.

Twin brothers and AI engineers Alberto and Alejandro Rosas founded Didit in 2023 with this future in mind, watching these market forces reshape identity.

Right now, generative AI does more than just talk and move and take actions. It also provides a method for fraudsters to pretend to be people or take actions faster than most people can react. Frontier models have made deepfakes, synthetic identities and injection attacks easier to create, weaponize and deploy than ever before.

At the same time, global age verification mandates are rushing ahead, alongside new financial compliance requirements that take into account the activity of AI agents. Fraud prevention requires attention to details about new types of AI-delegated activity and that means preserving digital identity that needs to go far beyond traditional Know Your Customer standards used by paperwork-bound businesses that exist today.

“No one was building for what was actually happening,” said founder and Chief Executive Alberto Rosas. “Fraud kept getting smarter, regulators kept getting stricter, and millions of new businesses suddenly needed to verify their users — but every existing provider couldn’t catch the new fraud, had painful onboarding and hid pricing behind a sales call.”

To capture this market, Rosas said, Didit leaned hard into identity using an application programming interface, providing developers the programmatic capability they needed to build it into their applications from the ground level — one API for identity, fraud and onboarding that provided pricing up front.

Addressing the shifting identity problem with AI

To get the job done, Didit connects to global government data sources to analyze more than 200 signals across verifications, including document authenticity, biometric liveliness, injection attack detection, deepfake analysis and behavioral signs.

The company builds its own AI models in-house with an aim to address the needs of the market and avoid third-party dependencies.

It also built its own specialized verification flows for every country, optimized around face types, skin tones, document formats and lighting conditions. All of these different factors open up various biases that can arise within AI models and datasets that need to be carefully curated. Otherwise they can lead to problematic reactions from anti-fraud detections.

At the end of the day, a human agent still needs to double-check red flags. But lowering the likelihood that people are falsely identified as fraudsters simply because they don’t “look” like what an AI model was trained on as a default person becomes important when identity, ethics and regulatory compliance are on the line.

“We started with identity verification because it was the hardest piece,” added Rosas. “What we’re really building is the trust layer for the internet.”

The company said that its long-term vision is to provide an identity wallet that will allow people to verify themselves once and then reuse that identity everywhere. That will allow developers to use a single API call — and also for AI agents, a single Model Context Protocol tool call — and consumers to make a single click wherever they go. It would become an internet passport that would open up any e-commerce site, government website or business for human or digital identities alike.

Didit made the throughline clear: It wants to make identity programmable in the same way Stripe Inc. and Twilio Inc. made payments and communications programmable.

To date, the company has already grown to more than 1,500 business to business customer across fintech, cryptocurrency, marketplaces, digital gaming, mobility and government. It added that the company serves customers in more than 220 countries and territories by bringing its service to a large number of customers that previously did not have an identity verification provider.

Photo: Truthout.org/photopin cc