TL;DR:
- EU officials want all AI-generated content to be clearly labeled to combat disinformation.
- Companies using generative AI tools should implement technology to recognize and label content that may spread false information.
- Safeguards need to be established by companies integrating generative AI into their services to prevent misuse.
- The EU’s 2022 Code of Practice on Disinformation has been signed by major tech companies, but Twitter’s withdrawal raises concerns.
- The forthcoming EU Artificial Intelligence Act will provide comprehensive guidelines for AI use.
- European officials urge companies to adopt a voluntary code of conduct for generative AI developers.
Main AI News:
In a bid to enhance transparency and combat the spread of disinformation, officials within the European Union (EU) are advocating for the labeling of all AI-generated content. Recognizing the potential of artificial intelligence tools, including OpenAI’s ChatGPT, to generate misleading information, the EU seeks to ensure that users are informed about the nature of the content they encounter.
Vera Jourova, Vice President for Values and Transparency at the European Commission, emphasized the need for companies utilizing generative AI tools to take responsibility for identifying and labeling content that could potentially disseminate disinformation. “Signatories offering services capable of disseminating AI-generated disinformation should implement technologies that can detect such content and clearly mark it for users,” Jourova stated.
Jourova further highlighted the importance of safeguarding measures for companies incorporating generative AI into their services. Notably, she mentioned Microsoft’s Bing Chat and Google’s Bard as examples and urged these companies, along with others, to establish safeguards that prevent malicious actors from exploiting AI for disinformation purposes.
In 2018, the EU introduced its “Code of Practice on Disinformation,” serving as an agreement and tool for the tech industry to uphold self-regulatory standards against disinformation. Notably, major tech companies such as Google, Microsoft, and Meta Platforms have already endorsed the EU’s 2022 Code of Practice on Disinformation. Jourova stressed that these companies, among others, should provide updates on the new safeguards they will implement for AI by July of this year.
Jourova also drew attention to Twitter’s withdrawal from the code of practice, cautioning the company to expect heightened scrutiny from regulators. “Twitter’s departure from the Code has attracted significant attention, and its adherence to EU law will face rigorous and urgent examination,” she remarked.
These statements from the EU Vice President come as the union prepares to introduce the EU Artificial Intelligence Act. This forthcoming legislation aims to establish comprehensive guidelines governing the public use of AI and the practices of companies deploying AI technology. While the official laws are expected to take effect in the next two to three years, European officials are encouraging companies to adopt a voluntary code of conduct for developers of generative AI in the interim.
Conclusion:
The EU’s push for transparent labeling of AI-generated content signals a significant development in the market. Companies utilizing generative AI tools will need to invest in technologies that can identify and clearly mark potentially misleading content. This move aims to enhance transparency and combat disinformation, ensuring that users are well-informed. The EU’s Code of Practice on Disinformation, endorsed by major tech players, sets the stage for self-regulatory standards.
However, Twitter’s departure from the code has raised concerns and may attract intensified regulatory scrutiny. As the EU prepares to introduce the Artificial Intelligence Act, businesses will face increasing pressure to establish safeguards and adhere to comprehensive guidelines governing the public use of AI. This evolving regulatory landscape emphasizes the need for responsible and accountable AI practices in the market.
