Brazil’s government wants to set new standards for the challenges posed by the social media environment and the increasing use of artificial intelligence with various draft laws in Congress.
On Thursday Senate leader Rodrigo Pacheco proposed a new framework for AI in the country, preceding a similar effort announced shortly thereafter by US President Joe Biden. The UK government has also just launched a first review of AI to assess its impact on consumers, businesses and the economy.
In line with some of the current global debates about emerging technologies, these issues are provoking heated debate.
“I think this is one of the most important issues these days. We are also paying close attention to the evolution of technology and business demand,” Cláudio Stopatto, general manager of Lenovo’s Infrastructure Solutions Group (ISG) in Brazil, told BNamerica’s growing use of AI.
Lenovo’s ISG is one of the leading manufacturers of servers for processing information in data centers. Some industry estimates suggest that new advanced AI tools will require 10x to 100x more processing power, with significant impacts on storage space and power consumption.
Lenovo is committed to developing AI and has invested in more powerful and compact servers, including for the Brazilian market, Stopatto said.
Beyond the question of infrastructure, some of the most pressing issues in advanced AI systems revolve around intellectual property, system bias, employment implications, privacy and data sovereignty. In Brazil, advanced AI systems like ChatGPT dominated the debate at the WebSummit tech event, held May 1-4 in Rio de Janeiro.
Senator Pacheco’s proposal is an attempt to address some of these issues.
The draft law is based on five key themes for AI: principles, data subject rights, risk classification, governance obligations and requirements, and oversight and accountability.
“By having governance tools and an institutional structure for inspection and monitoring, [the bill] creates predictable conditions for its interpretation and ultimately legal certainty for innovation and technological development,” argued Pacheco.
The proposal is the result of recommendations from a working group established in 2022 and brings together various bills previously tabled in Congress into a single legal framework for AI.
The text aims to establish general standards for the “responsible development” and implementation of “secure systems for technological tools in Brazil” with the aim of protecting fundamental rights, according to the explanatory memorandum to the proposal.
Among other things, it states that companies must offer up-front information about interactions with AI systems and the explanation of decisions or recommendations they make, giving the user the right to challenge them and request human intervention if necessary.
The text also states that individuals cannot be discriminated against by AI systems, a sensible point if vague in practical terms.
Pacheco’s proposal also stipulates that AI vendors or operators that cause material, moral, individual or collective harm will be required to fully repair them “regardless of the degree of autonomy of the system”.
It also outlines the definition of an AI regulator linked to the executive branch of government, as well as corporate accountability. No date has yet been set for voting on the proposal.
The widespread use of AI has raised concerns about prejudice and the replacement of human labor with machines, although the labor issue is not directly addressed in Pacheco’s bill.
According to IDC, in 2027, around 50% of the top 5,000 companies in Latin America “will continue to face significant risks because frontline workers and business leaders are unwilling to trust actions initiated by proven autonomous technology systems.”
IBM CEO Arvind Krishna has also estimated that 30% of non-customer-facing roles could be eliminated in the next five years.
The executive said IBM would freeze hiring as it expects to replace about 7,800 jobs with AI in the coming years.
The local debate about social media is far more controversial than that about AI, at least for now.
After intense pressure from Google and strong opposition from a number of evangelical lawmakers, the so-called “fake news bill” due for a vote this week has been postponed by Brazil’s lower house speaker, Artur Lira, pending greater consensus.
The text was already passed by the Senate in 2020, but subsequently received various changes in the lower house.
Opponents of the bill have generally argued against what they call “censorship” and the risks to free speech. Proponents allude to the need for greater accountability, transparency and liability for social media platforms and content moderation in the face of messages supporting hate speech or violence.
“The project outlines an important mechanism to create more transparency and user control over how content is moderated by the platforms,” said Renata Mielli, coordinator of the Brazilian internet governing body CGI.br, in an article.
“Due process establishes the user’s right to challenge the removal of posts, to receive reasonable justification for the decision, and to set a time limit for the platform to analyze the request for review of improper moderation, which protects freedom of expression,” said he.
According to lawyer Alexander Coelho, a specialist in digital law and data protection, the country’s existing legal system already has the necessary tools to solve some of the highlighted problems.
He said the country’s Internet accounting framework and other legal provisions are capable of preventing the spread of harmful content.
“All of this also contributes to the self-regulation of application providers,” he added in a statement. Coelho is a member of the Data Protection Committee of the São Paulo Bar Association OAB.
Organizations from the technology sector, such as the ICT association Brasscom and the electrical electronics association Abinee, emphasized in a joint letter their concerns about “legal certainty and harmonization with existing standards”, especially with regard to personal data.
“Regardless of the important debate on new rules to combat the spread of false information on the Internet, which is a global problem, there is a need to mitigate the risk of articles appearing uncertain during the long and mature debate that has led to the could evoke approval of the LGPD [data protection legislation] and the formation and powers of the ANPD [national data protection oversight agency]’ the letter said.
The digital advertising industry is also opposed to the project, particularly with regard to payments for news content shared on social media and new advertising rules.
“We have access to unofficial texts that concern us because they contain, among other things, definitions on programmatic media, new commercial responsibilities, rules for foreign companies and companies affecting the operations of many of our employees and the advertising market in general,” said the Digital advertising industry group IAB Brasil.
The telecom infrastructure sector, on the other hand, is more concerned with tax issues involving large technology companies that are not directly addressed in the text of the bill.
“Regardless of the discussion of content regulation, it’s important to remember that Big Tech pays far less tax – if any – than it should, compared to the telecom industry, which has to meet a number of regulatory requirements,” Vivien Suruagy, President of the call center and ICT infrastructure association Feninfra said in a statement.