Regulations, privacy and personal data in Artificial Intelligence

22/04/2021 22

EU Proposal about the limitation of indiscriminate surveillance and prohibition of social scoring, using AI.

The European legislators, as well as AI companies working under the requirements of data protection legislation, such as Sherpa.ai, are worried about AI-based scoring systems managed by governments, like the ones developed and used in China and about the uncontrolled use of personal data from consumers, contravening regulations like GDPR.

An extreme case is the usage of AI tools by the Chinese government, to identify demonstrators supporting democracy in Hong Kong, and to control and discriminate the Uygur Muslim community. Facial scanners are already used in China to identify and fine pedestrians, and residents in Shanghai will soon need to verify their identity in pharmacies using facial scanners as well.

AI adoption rules must be centered on the people

At Sherpa.ai we are fully aligned with the European proposal, and we must be at the forefront of the adoption of these rules. This is what is being debated in the European Parliament, the foundation of how the digital future of Europe will be shaped, and of the usage of AI in particular.

There are no doubts that AI is transformative and positive. It can provide answers to many social and economic challenges, from the treatment of illnesses to the minimization of the environmental impact of agriculture.

At the same time, it requires a high level of ethical standards and legal certainty for developers and users. For AI to be successful in Europe, consumers must trust in it, and regulations will help to build such trust, creating a favorable environment for investment and innovation for AI across all sectors.

Xabi Uribe-Etxebarria, Founder & CEO of Sherpa.ai:

“It is critical for organizations adopting AI and for emerging vendors like Sherpa.ai to create a common framework for all, to leverage the whole potential of AI while putting consumers’ rights in the EU at the center of our efforts”, according to Xabi Uribe-Etxebarria, Founder & CEO of Sherpa.ai.

The privacy of personal data must always be respected.

The main risks related to AI refer to the impact on consumers’ fundamental rights, particularly those related to the protection of their personal data and their privacy.

In fact, irrespective of the new EU proposal and regulations which may be enacted in the coming months or years, both developers and organizations offering AI solutions are ALREADY subject to the European legislation regarding data protection and privacy, consumer protection, safety regulations and product liability.

Logically, consumers expect the same level of security and respect for their rights irrespective of whether a product or system uses AI or not. However, some specific characteristics of AI, like opacity for example, may make it more difficult to ensure compliance with regulations. For this reason, it becomes necessary to examine if the current legislation can address the risks of AI and if it can be applied effectively, or if adaptions or new legislation are required.

Sherpa.ai proposes to ensure personal data privacy, by design.

In fact, at Sherpa.ai, privacy is one of the main areas of research and development, and a framework is already available for customers.

The AI Privacy Framework allows to train advanced AI models with complementary or different sources of information, for example, among organizations from different sectors, with different data types and sets, or among decentralized organizations with the same data types but decentralized data sets.

This type of solutions have already been used, among others, by the medical research community, for predictions in diagnosis, with different hospitals collaborating to create an AI model, each with its own data set, and always respecting the privacy of their patients.

At Sherpa.ai we are convinced that these new solutions will enable new collaborative business models across a variety of sectors. With our AI Privacy Framework, new advanced AI models can be generated without sharing any personal data, which never leave the infrastructure of participating organizations.

In this regard, we believe this EU proposal will boost opportunities for Sherpa.ai with its AI platform and AI Privacy framework, which were designed precisely to respect users’ privacy and to comply with privacy regulations.