Brazil’s AI Law Faces Uncertain Future as Big Tech Warms to Trump

Laís Martins is a Brazilian journalist based in São Paulo and a fellow at Tech Policy Press.

Over the past two years, Brazilian lawmakers have debated how to regulate artificial intelligence. After months of discussion, the Senate voted on a bill in December that is now headed to the lower house of Congress. But the world, and particularly the tech landscape in which the Senate passed the bill, was quite different just two months ago.

In the United States, Donald Trump had not yet taken office, had not revoked a 2023 Executive Order on AI by former president Biden, and had not put in its place an order with the primary goal to “sustain and enhance America’s global AI dominance.” And more significantly, Silicon Valley leaders like Meta founder and CEO Mark Zuckerberg had not yet acquiesced to the incoming administration’s vision for tech and the US, at least not publicly. These events, which took place in just the past few weeks, could significantly impact the prospects of Brazil’s AI bill ever making it into law.

The Senate’s ‘Watered Down’ AI Framework

Brazil began its efforts to regulate AI in February 2022, when Senate President Rodrigo Pacheco formed a committee of legal experts tasked with drafting a report to help guide lawmakers' development of AI legislation. In December of that year, the experts delivered to Pacheco a report recommending a risk-based approach to AI regulation.

Many of the elements in the experts’ report were then incorporated in a bill proposed by Pacheco in May 2023. To review the draft bill, the senators created a temporary commission for artificial intelligence that, over 19 months, discussed drafts, held 24 meetings and 14 public hearings, and gathered input from dozens of outside experts.

In December 2024, the commission approved a watered-down version of the bill, which the Senate passed on the floor the next day. Still, Data Privacy Brazil, an organization focused on data protection and digital rights, provided that “The legislation remains protective of fundamental rights and advances crucial standards for a fair informational ecosystem in the coming years.”

The legislation favors a risk-based regulatory model, imposing obligations on developers, distributors, and applicators of high-risk systems. Under this regime, these entities would have to produce risk assessments that address biases and potential discrimination, as well as create channels for individuals affected by automated decisions to appeal to human reviewers.

The regulatory framework classifies certain applications –such as traffic control, student admissions, hiring and promoting employees, and border and immigration control as high-risk. However, lobbying pressure from the tech industry led to several key carveouts. For example, algorithms that curate, recommend, and distribute social media content were removed from the high-risk category, shielding platforms from heightened scrutiny and transparency requirements.

The bill’s most distinctive feature is perhaps a provision that mandates companies to compensate copyright holders when their content or data is used to train AI systems. It also creates an exception, allowing academic and other public institutions to use third-party content to train AI systems for research purposes.

Beyond removing recommendation systems from the high-risk list, other notable ommissions appeared in the final version that cleared the Senate. Lawmakers removed most references to “information integrity” or “freedom of expression” after pressure from big tech companies and opposition members of Congress who claimed such provisions could give margin to “censorship.”

The Brazilian bill shares some similarities with the European Union’s AI Act, approved in March 2024. Both adopt a risk-based regulatory model structured around different tiers of risks. However, an analysis by think-tank ITS Rio de Janeiro, based on the November 2024 version of the Brazilian bill, found that it imposes broader obligations than the EU legislation. Unlike the EU’s approach, which places specific responsibilities on a select group of actors, Brazil’s bill applies obligations to more actors in the AI supply chain. More than half of the 56 obligations laid out in the bill apply equally to providers, applicators, and distributors, whereas in the EU’s AI Act, the majority of the obligations apply only to the developers of the technology. It also identified that the Brazilian bill has obligations specifically about the use of AI in the public sector, while the EU does not mention this use.

Some experts have raised concerns about the inadequacy of the EU law in addressing “harms that have already been caused by the rapid proliferation of these technologies.” They have also criticized the tiered timeline of implementation - replicated in the Brazilian legislation - arguing by the time the regulations take effect, it may be too late to mitigate the negative impacts of AI.

Does Big Tech’s Lobbying Playbook Change?

While the AI bill faced predictable opposition from right-wing politicians who were quick to apply the ‘censorship label’ to it, even though it does not regulate content nor discourse, lobbying from technology companies did not follow that script.

Indeed, tech companies took a notably different approach compared to their actions related to the failed “Fake News” bill, a comprehensive law to regulate platforms that made its way through Congress starting in 2020 but was ultimately shelved in 2024. During the debate over the bill, big tech launched an aggressive campaign to block it. They lobbied Congress members and sought public support to stop the vote on the legislation, using problematic tactics. For example, Google exhibited a banner on its Brazilian homepage in which it claimed the bill would “sow confusion about what is true or not,” and Meta bought a full-page ad in Brazil’s most prominent newspaper warning that the bill ‘would hurt small businesses.’ Google’s action led to a Federal Police investigation into its potential abuse of economic power, though officials later dismissed the case.

But big tech firms' reputations were tarnished, leading to a change of strategy when it came to the AI bill. This time, the companies did not come out publicly against the bill. Instead, their arguments against the bill appeared in the voices of senators sympathetic to a more lenient regulatory approach or in the remarks of the temporary AI commission’s outside experts who were connected to organizations funded by big tech. In March 2024, a Brazilian delegation, including senators and the vice president of the temporary AI Commission, traveled to Washington DC and met with officials of the US government, think-tank representatives, and executives from Amazon, Google, and Microsoft. When they returned from DC, two senators who had attended the trip submitted dozens of amendments, some written in identical language, attempting to weaken the bill.

Advocates for a weaker regulatory approach (or no regulation at all) were joined by Brazilian industry groups and data center executives, all echoing a big tech narrative that regulation would hinder innovation. To some extent, their efforts were successful in weakening what started out as a very ambitious bill.

As Brazil’s AI bill now heads to the Lower House of Congress, civil society groups already expected even greater pressure from oppositional politicians and the private sector, particularly big tech. With Trump now again in power, many Silicon Valley billionaires have fully and publicly embraced the Trump presidency. In addition to Elon Musk’s prominent support of Trump, Mark Zuckerberg has vowed Meta will return to its roots and restore “free expression,” announcing that the company will “work with President Trump to push back on governments around the world” who are “going after American companies and pushing to censor more.”

The coming months will reveal big tech’s new emboldened approach to resisting regulation. Brazil’s experience may serve as a key example of how far US tech giants, now backed by the US government, will go to shape global AI regulation in their favor.