SIGEN's AI Control Guide — Argentina's National Audit Office Sets Lifecycle Rules for Public-Sector AI

On 22 June 2026, Argentina's national internal-control body, the Sindicatura General de la Nacion (SIGEN), published Resolution SIGEN 197/2026 in the Boletin Oficial, creating the country's first government-wide control framework for artificial intelligence in the National Public Sector. SIGEN says the guide followed its own survey of state agencies, which found a range of AI projects already under way -- from staff using generative tools informally to AI embedded in institutional systems -- without adequately implemented controls.

The guide, presented publicly on 11 July 2026 by SIGEN head Alejandro Diaz, distinguishes two forms of use: individual use of generative AI tools by public employees, and institutional use, where AI is built into an agency's own systems. For both, it requires an explicit institutional policy, staff training on responsible use, data-governance controls covering quality, integrity and security, controls spanning the full system lifecycle from feasibility analysis to production, periodic performance reviews, defined roles and access limits, complete documentation for audit purposes, and transparency measures toward citizens.

It is explicitly framed as a preventive, risk-based approach rather than a ban, and names risks such as unauthorised disclosure of confidential information, mishandling of personal data, dependency on external AI vendors, absent validation criteria for AI-generated outputs, and staff unfamiliarity with model bias and vulnerabilities.

As a newly issued control framework rather than a deployed AI system, the guide has no operating results yet: SIGEN has not published data on how many of the surveyed agencies have since implemented the required controls, and independent audits of compliance have not yet been reported.