TSE advances on big techs in the election amid obstacles – 01/12/2024 – Power

With the release of a draft resolution whose main objective is to regulate the use of artificial intelligence in this year’s municipal elections, the TSE (Superior Electoral Court) seeks to increase the accountability of big techs for content published on social networks.

A key point of the court’s proposal, which must still be approved in the court’s plenary session, is to assign responsibilities to so-called “internet application providers” that broadcast electoral content.

They must, according to the text, adopt and publicize “measures to prevent or reduce the circulation of illicit content that affects the integrity of the electoral process.”

Among these measures are guarantees of “effective notification mechanisms, access to a reporting channel and corrective and preventive actions”.

The draft also establishes rules regarding other issues related to electoral propaganda on the internet, such as the promotion of content by companies and also electoral lives.

What the TSE has highlighted in the text is, above all, the obligation for electoral propaganda to explicitly inform the use of content manufactured or manipulated through digital technologies.

The proposals come as a result of gaps in the Legislature regarding the issue, despite several demands made by the ministers of the electoral court in the last year.

In a series of speeches, the president of the court, Alexandre de Moraes, defended punishments for the use of artificial intelligence to manipulate elections.

“Manipulated the voter, won the election, fine. The sanction must be drastic. Whoever uses artificial intelligence to manipulate the voter’s will to win the elections, if discovered, revocation of registration, and if elected, revocation of mandate “, he said in December.

At the time, Moraes reinforced a request for regulation of the tool by the National Congress.

He also said that, despite being a technological advance, artificial intelligence can be distorted by big techs and, therefore, they must be held responsible.

The absence of this regulation is seen by observers as one of the reasons for the court to propose a resolution on the issue.

“In the 2024 electoral cycle we did not have a reform approved by Congress, the proposal only passed through the Chamber of Deputies, however it did not go forward in the Senate”, says Edson Borowski, a civil servant at the Electoral Court and member of Abradep (Brazilian Academy of Electoral and Political).

“It is important to draw attention to the fact that internet spaces (social media and messaging applications) have been the main focus of intervention by the Electoral Court, at least since the 2018 elections”, he states.

The text will still be debated in a public hearing on January 25th, before going to the court plenary, but it already divides experts. The questions range from a lack of clarity to obstacles to compliance with the rules.

“A duty of transparency is created on internet application providers in the provision of content boosting services, even when verified in a period prior to the electoral campaigns”, says Alexandre Freire Pimentel, judge at the TJ-PE (Court of Justice of Pernambuco) and also a member of Abradep.

However, despite transparency obligations regarding advertising and amounts paid, Pimentel says that there are no effective technological means to determine the real size of the boosts.

“The systems used by providers are opaque neural networks and their source codes are protected by the guarantee of industrial and commercial secrecy”, he states. “The law exempts providers from liability when they are unable to comply with court orders due to technical impossibilities in their systems.”

“Thus, the effectiveness of this rule will depend, mainly, on the cooperation of big techs”, adds Pimentel.

According to lawyer Raphael de Matos Cardoso, a doctor in State law, “it is not clear how the use of artificial intelligence will be controlled, since Brazil still has no regulation in this regard.”

Lawyer specializing in data protection and artificial intelligence Daniel Becker questions the liability of companies if incorrect information is published.

“Putting the burden on providers of providing distorted information — which is not a direct product of their actions — is an unfair choice, and one that relieves the person truly responsible for the illicit act,” says Becker.

As it has not yet been approved, the text of the draft can still be modified and is open to suggestions that can be sent to the TSE website until January 19th.