Brazil Ahead of the US and the Global Average

A global Sinch study, conducted with 2,527 executives and published throughout the week of August 13 to 17, 2026 by outlets including TI Inside, Olhar Digital and SitePD, shows 76% of Brazilian companies keep artificial intelligence agents in production. That figure beats the 62% global average and also tops the 67% recorded in the United States, placing Brazil among the world's most advanced countries in practical, not just experimental, adoption of corporate AI agents.

The Flip Side of the Lead: 80% of Companies Have Already Pulled Back

The same survey brings a direct counterpoint to the good news: 80% of Brazilian organizations have paused or reversed some AI agent deployment at some point. This is not abandonment of the technology, but a recurring pattern of trial, error and correction that tracks the speed at which the country has been putting autonomous agents to work inside real business processes.

Data Leaks Are the Most Common Reason for Pulling Back

Among the cases where a company had to pull back, 39% involved a data or personal information leak, according to the research. That figure is significantly higher than what was observed in North America, where 26% of organizations cited the same reason. More specifically, 41% of Brazilian organizations reported having to fully shut down an AI agent after a leak, versus 26% in North America.

Companies Now Allocate More Resources to Security Than to Development

One data point signals a shift in priorities: 76% of Brazilian companies now allocate specific resources to trust, security and compliance around their AI agents, surpassing the 63% investing directly in developing new AI capabilities. In other words, shielding what is already in production has started competing, in investment terms, with the race to launch more agents.

A Pattern of Growth and Pullback That Keeps Repeating

Related reports published in the same window, including a Netskope analysis on data breaches tied to AI agents, and an earlier study from May that already found 74% of companies had pulled back on agent use after failures, reinforce that Brazil is living a consistent cycle: fast adoption, security incidents, a targeted pullback, and then another attempt, usually with tighter controls.

Why It Matters for Brazilian Agencies and SMBs

For marketing and customer service automation agencies that have already deployed or plan to deploy AI agents to handle customer data, the Sinch study works as a direct practical warning: the odds of having to shut down an agent over a data leak in Brazil are nearly double the North American average. Before scaling an autonomous agent into production, it is worth locking down three minimums: a tightly restricted scope of data the agent can access, a complete log of every automated action taken on personal data, and a fast shutdown plan tested in advance, not improvised in the middle of an incident. Given that 76% of companies already treat security as an investment priority equal to development, agencies that have not done this yet risk falling behind, not on AI adoption, but on the governance maturity the market itself now demands.