Tools change fast. What makes the difference is the right people, and the right infrastructure!
Deepfakes, synthetic documents, and ready-made fraud kits. None of this is new. What changed is access. Today, anyone can generate a fake identity with consistent metadata or tamper with a digital document without any technical skill. In Brazil alone, scams using AI-manipulated audio and video grew 267% between July 2023 and July 2024, according to Kaspersky. When that capability meets high financial incentives, the result is what the market is now seeing: a jump in sophistication, volume, and speed across fraud attempts.
Jason Howard, CEO of Certta, points to this shift as the biggest inflection point in digital security over the past few years. No single technology changed the landscape on its own. It was the combination of accelerated digitalization with tools once confined to labs or highly specialized fraudsters that, once they became accessible, redistributed risk across the entire chain.
Digital Fraud in Brazil: Why the Country Became a Defense Laboratory Against AI Attacks
The Fraud as a Service model consolidated what used to be a fragmented market. Underground forums now offer ready-made phishing kits, leaked databases, bot infrastructure, and highly persuasive document templates, many of them powered by generative AI. The practical effect is the industrialization of attacks: more actors, more variations, and an update cycle that puts pressure on operations still reacting instead of anticipating.
For Jason, accepting this shift as structural is the first step. “You can’t treat digital fraud as a one-off risk, because it’s structural to the business,” he says. The second step comes after that acceptance: choosing the right partners, designing a decision architecture, and building response capacity that doesn’t rely on improvisation.
“No company, on its own, can keep up with the pace of innovation in today’s fraud market. It’s essential to work with vendors who are on the front line of deepfake detection, document analysis, and threat intelligence.” — Jason Howard, CEO of Certta
Brazil as a Defense Laboratory
Brazil combines a set of conditions few markets share: high technology adoption, substantial transaction volume, and a fraud ecosystem that tests new tools at the same pace the legitimate market innovates. Pix, which brought digital payments to nearly the entire population, also expanded the available attack surface.
But there’s a side to this picture that risk coverage doesn’t always capture: Brazil has also built a sophisticated defense ecosystem. Jason notes that the country leads on several fronts in digital payments and, in his view, in the ability to defend against fraud as well. Constant pressure has created an environment where security innovation happens out of necessity, not choice.
AI in Three Phases: From Authentication to Autonomous Agents
Jason breaks AI adoption in security down into three phases. The first is nearly mandatory: any operation that receives digital documents, images, or unstructured information needs AI to verify authenticity. Payment slips, contracts, identity documents, the technology to detect tampering already exists and works. The problem is that few companies have implemented it consistently.
In practice, a crypto exchange that collects identity documents during onboarding is already in phase one when it uses AI to check whether the photo is real, whether the document has been tampered with, and whether the metadata matches the expected pattern. Operations that still handle this verification manually, or rely on fixed rules, are operating below the minimum bar the current landscape demands.
The second phase is operational efficiency: AI with real capacity to automate business registration, contract analysis, and corporate document checks, freeing up teams for higher-value work and cutting turnaround times substantially.
A financial institution that uses AI to process business account openings, for example, no longer depends on analysts to review every corporate document by hand. What used to take days now gets resolved in minutes, with a smaller margin of error and full traceability of every decision.
The third phase, now arriving, involves AI agents: systems that execute tasks end to end. The challenge at this stage won’t be access to technology but curation, having the right people to filter and integrate the options emerging in the market, embedding AI into a larger architecture with business rules, monitoring, and governance.
In this phase, an agent can run the entire risk analysis flow for a transaction: gather data, trigger the relevant verification layers, log the decision, and escalate ambiguous cases for human review, all without manual intervention along the way. The differentiator won’t be the agent itself, but whoever designed the rules that govern it.
“It would be easy to eliminate fraud if you simply shut the doors, but that’s not the goal.” — Jason Howard
Verification Intelligence Hub: An Infrastructure Decision, Not a Tool Decision
When Jason talks about intelligent verification, he means assembling a set of tools, integrating data, and developing technologies to make more precise risk decisions without slowing down the journey. Certta’s Verification Intelligence Hub applies that principle: low-value transactions move forward with less friction, while higher-risk ones trigger specific layers, such as biometrics, advanced document verification, or checks against specialized databases.
There’s another component Jason sees as essential for the next cycle: collaboration between companies. Fraudsters work as a network and exchange information in real time. The industry needs to evolve toward sharing intelligence without exposing sensitive data. The technology to enable that kind of exchange already exists. Once the market moves on this, the impact on prevention will be substantial.
What holds this back today isn’t a lack of willingness. It’s a mix of regulatory caution, competitive pressure within the sector, and the absence of shared standards. Companies that could alert one another about the same CPF (Brazil’s individual taxpayer ID) being used in simultaneous fraud attempts choose to keep that information internal, out of concern over violating Brazil’s data protection law (LGPD) or handing a competitor an advantage. The result is that each company redoes, on its own, work that could be distributed across the industry.
Certta operates at exactly that convergence point. As a hub that integrates multiple data sources and serves companies across different sectors, the platform builds up cross-industry intelligence that no single company could develop on its own. Risk signals that surface across multiple operations simultaneously feed models that improve decision accuracy for every participant, without any company having to expose its customer base or share sensitive data directly with competitors.
Digital Trust as a Strategic Asset: Building the Foundation Early Means Higher Approval Rates and Controlled Growth
Jason closes with an idea that runs through the entire conversation: digital security is becoming an enabler of new business models, not just a barrier. Whatever device or channel dominates in the years ahead, the need to build trust continuously, combining technology, governance, and collaboration across the whole ecosystem, remains constant.
Companies connected to the right data, the right partner ecosystem, and a robust verification architecture will be better positioned to navigate this period. For Jason, whoever builds that foundation early doesn’t just protect itself better. It approves more, grows with control, and sustains the revenue it has already earned.
“Tools will keep changing very quickly, but what will really make the difference are the right people, connected to the right data and the right partners.” — Jason Howard
