Generative AI Models May Have Unlawful Origins, Regulator Warns

By Netvora Tech News

The Authority for Personal Data (AP) in the Netherlands has raised concerns that many generative AI models may have unlawful origins due to the use of massive amounts of scraped information and user-input data. The regulator presented a vision for responsible development and deployment of generative AI at a recent presentation. According to the AP, the overall assessment is that the majority of current generative AI models fall short in terms of lawfulness. This is because nearly all publically available data on the internet has been scraped for training foundation models, which are used in AI applications. It is likely that these data sets also include illegally collected sensitive personal data. However, the AP notes that the further use of these foundation models by Dutch and European parties is not necessarily unlawful, as determined by an analysis by the European Data Protection Board (EDPB). The AP aims to provide organizations with more clarity on the use of generative AI and the conditions that must be met. "Generative AI has the potential to surprise. The technology can bring many benefits, but also pose concerns," said AP chairman Aleid Wolfsen. "It is essential that the protection of personal data is also in order when using generative AI. Given the rapid technological developments, correcting mistakes afterwards is not sufficient."

Concerns Over Data Scraping

The AP is concerned about the large-scale scraping of data from the internet, which is often used to train AI models. This data includes not only publicly available information but also user-input data, which can contain sensitive personal information.

The AP emphasizes that the protection of personal data is essential, especially in the context of AI development. The regulator is working to provide organizations with clear guidelines on the use of generative AI and the conditions that must be met to ensure compliance with data protection regulations.

Foundation Models and Data Protection

Foundation models are used in generative AI applications to analyze and generate data. The AP notes that these models are often trained using massive amounts of data, which are scraped from the internet. This raises concerns about the potential inclusion of illegally collected sensitive personal data in these models.

The AP is working to develop guidelines for the use of foundation models and other AI technologies, with a focus on ensuring the protection of personal data. The regulator is also working with other European authorities to develop harmonized guidelines for the use of AI technologies.

Conclusion

The AP's concerns highlight the need for responsible development and deployment of generative AI. As the technology continues to evolve, it is essential that regulators and organizations work together to ensure that personal data is protected and that AI technologies are used in a way that benefits society.

The AP's vision for responsible development and deployment of generative AI is an important step towards achieving this goal. By providing clear guidelines and conditions for the use of AI technologies, the AP aims to ensure that the benefits of AI are realized while minimizing the risks.

The AP has raised concerns about the unlawful origins of many generative AI models.
The regulator is working to develop guidelines for the use of generative AI and foundation models.
The AP emphasizes the importance of protecting personal data in the context of AI development.