Brazilian associations notify AWS, Apple, Google, Meta, Microsoft, OpenAI over use of protected works to train artificial intelligence
Associations representing newspapers, magazines, radio and television broadcasters, as well as the music and copyright industries, have formally notified artificial intelligence (AI) companies operating in Brazil in an effort to negotiate authorization for the use of copyrighted content in AI tools, the entities said in a statement on Thursday (19).
The companies notified include the Brazilian subsidiaries of Amazon Web Services (AWS), Apple, Google, Meta, Microsoft, and OpenAI, according to the National Association of Newspapers (ANJ).
In a joint statement, ANJ, the Brazilian Association of Radio and Television Broadcasters (Abert) and the National Association of Magazine Publishers (Aner) said that “authorization for the use of protected content in AI tools is guaranteed by law” and that the purpose of contacting AI companies “is to build a bridge between technology and copyright holders, ensuring that the advancement of AI in Brazil occurs in harmony with the sustainability of those who produce quality information and culture.”
“The notification is a call for dialogue and negotiation, as well as for respect for existing laws,” ANJ executive president Marcelo Rech said in the statement. He added that “media outlets support the development of AI, but this cannot become a lawless territory. Like any responsible economic activity, AI developers must follow minimum rules recognizing intellectual property, which includes journalistic content.”
The notifications were sent on December 22 to Brazilian subsidiaries of multinational AI companies. The letter was signed by Abert, ANJ, Aner, the Brazilian Association of Music and Arts (Abramus), the Association of Musicians, Arrangers and Conductors-Brazilian Musical Society (Amar/Sombrás), and the Association of Performers and Musicians (Assim).
According to Rech, journalistic information plays a decisive role in AI systems. “It is estimated that around one-third of the digital content that has fed and sustained AI comes from journalistic outlets,” he said in the statement. “It is therefore only fair and natural that this third-party production, which has a cost, be compensated by another commercial activity that makes use of it.”
Rech also noted that the signatory entities have consistently defended negotiation and agreements as the basis for coexistence and development between journalism and AI. “In the absence of prior authorization for the use of journalistic content, however, it is legitimate for media outlets to seek their rights in court, as has already been happening in many countries, including Brazil,” he said.
“The organizations emphasize that the proposal extends to all platforms and developers of AI systems that use or intend to use protected content produced by their members,” the statement said. “The sector is open to negotiating authorization, licensing, and partnership models that ensure legal certainty and mutual benefits, safeguarding sustainable growth and investment in Brazil’s creative ecosystem.”
In the letter sent in December, the entities stated that they understand “that the advancement of AI must respect the copyrights and intellectual property of content produced by our members, as provided for in the Federal Constitution and Law No. 9.610/98 [Brazil’s Copyright Law].” They warned that unauthorized use of such content could undermine the ecosystem of journalistic and artistic production, discourage intellectual creation, and, above all, violate rights.
The letter continues: “Therefore, should there be interest in using content from our members for purposes including, but not limited to, data mining, training, or development of AI systems, we are available to discuss forms of authorization, compensation, and partnerships that benefit all parties involved and ensure the protection of copyright over such content.”
“If such use is already taking place, we request that you contact the Associations and/or the affiliated outlet whose content was used, so that an amicable solution may be negotiated to avoid future litigation,” the letter adds.
The entities conclude by reiterating that their “objective is to promote dialogue and seek innovative solutions, always respecting copyright and current legislation,” and that they remain open “to meetings that may result in agreements beneficial to all.”
As of Thursday (19), the associations had not received responses from the companies, Rech told Valor.
The use of copyrighted news content to train AI tools has been the subject of a wave of lawsuits and licensing agreements between media companies and major technology firms worldwide. A survey by the Tow Center for Digital Journalism at Columbia University lists 160 lawsuits and agreements between news organizations and AI companies since June 2023. In February alone, up to Thursday (19), agreements were announced between Vox Media and Hearst Magazines with Microsoft, between Axios and OpenAI, and between the Financial Times and Google.
In Brazil, Rech cited a licensing agreement signed in December between newspaper O Estado de S. Paulo and Google to train the Gemini AI tool. “We believe this is the path that should be followed by Google and all other developers [of AI tools] that rely on journalistic content in their activities,” he said.
ANJ and other sector associations have also been active before Brazil’s antitrust authority, the Administrative Council for Economic Defense (CADE), in an inquiry investigating possible abuse of dominant position by Google in search and alleged misuse of journalistic content in Brazil. In late August, CADE decided to deepen its investigation into the effects of algorithm changes and AI tools on news traffic.
Blocking access by AI tools has been one response adopted by media companies. Valor columnist Guilherme Ravache cited a survey by public relations platform BuzzStream showing that 79% of nearly 100 leading news websites in the United Kingdom and the United States already block at least one crawler used for AI training by OpenAI, Anthropic, Apple, and Google.
Contacted by Valor, Apple, Google, and Meta declined to comment. AWS, Microsoft, and OpenAI did not respond.
This article was translated from Valor Econômico using an artificial intelligence tool under the supervision of the Valor International editorial team to ensure accuracy, clarity, and adherence to our editorial standards. Read our Editorial Principles.
