Zetta lança Guia de Inteligência Artificial Generativa para o mercado financeiro e de pagamentos

Zetta, an association that represents the largest fintechs in the country, with institutional support from Anbima (Brazilian Association of Financial and Capital Markets Entities), has just launched the Best Practices Guide for the Evaluation of Generative AI Providers (GAIAG). The document is the first technical guide dedicated exclusively to the financial and payments market in the country, and should act as an accelerator for the safe adoption of Generative Artificial Intelligence (GenAI), structured in five pillars: compliance, cybersecurity, AI life cycle, data protection and value chain.

“O launch shows Zetta's commitment to supporting Brazil's leadership in the digital economy, reinforcing that responsible adoption of generative AI has the potential to expand access to financial services, strengthen anti-fraud mechanisms and improve the user experience”, says Fernanda Laranja, vice president of Zetta.

The need for the guide arises from the mismatch between rapid adoption of the technology and the absence of clear governance standards, in a scenario where the accelerated incorporation of Generative AI into the financial sector brings significant operational, ethical and regulatory challenges. 

The central objective is to establish a Brazilian financial and payments industry standard for third-party risk qualification that provides Generative Artificial Intelligence (IAGen) solutions. “O GAIAG is not only a theoretical document, but a practical and modular tool designed to promote a responsible innovation environment, ensuring safety and operational excellence”, says Samanta Oliveira, coordinator of the Zetta AI and data working group. 

“O Guide reflects a joint standardization effort to strengthen the innovation ecosystem in the financial sector. It reaffirms Anbima's commitment to promote and support digital transformation initiatives in the investment industry and capital market, supporting institutions in the responsible adoption of AI. By promoting consistent content and technical references, we create conditions for the market to use technology to generate business value with security, governance and” efficiency, says Carlos Andre, president of Anbima.

The initiative, which also had technical support from the law firm Serur and the consultancy Alvarez & Marsal, sets a framework for the safe adoption of Generative AI in Brazil, proposing a robust framework for the evaluation and hiring of tools.

“O Guia is a fundamental step towards the maturity of the Brazilian financial innovation ecosystem. By technically supporting the construction of this framework, our goal was to create an instrument that allows companies to evaluate and hire Generative AI solutions with security, efficiency and alignment with the best global governance practices”, says Fabricio da Mota Alves, partner at Serur.

Developed collaboratively, the document establishes technical standards and encourages the exchange of best practices that raise the regulatory and technological maturity of the ecosystem, aligning it with international references and anticipating debates that are already advancing in the National Congress and regulatory bodies.

“The consolidation of best practices for evaluation of generative artificial intelligence providers is essential to strengthen governance, mitigate risks and ensure regulatory compliance. By contributing to this guide, we reinforce a structured approach that supports safer and more informed decisions in the adoption of AI in Brazil”, says Eduardo Magalhaes, Managing Director of Disputes & Investigations at Alvarez & Marsal. 

The GAIAG works through a structured methodology in stages, supported by a practical tool (evaluation worksheet) and guided by the principle of risk-based proportionality. Before analyzing the supplier, the contracting institution evaluates the context of use of AI, classifying the risk in four levels (minimum, limited, high and critical) from criteria such as purpose of use (if it is only administrative support or end-activity), degree of regulatory complexity and need for licenses of the contractor, operational impact in case of unavailability of the solution and sensitivity of the data involved. Then, the evaluation of the provider (due diligence is carried out after a risk-related risk-related risk is generated by the analysis of the risk-related risk-related risk-related to the contract.

Access the Best Practices Guide for Generative AI Provider Assessment (GAIAG) at the link:  https://gaig.somoszetta.org.br/