Continued from the previous page
These fines range from 35 million euros (or 7 per cent of turnover) for the most serious violations, relating to systems with unacceptable risk, to 15 million euros (or 3 per cent of turnover) for violations relating to data security and management, e.g. failure to provide documentation and information to the authorities, to 7.5 million euros (or 1 per cent of turnover) for providing inaccurate or misleading information.
In the light of this Regulation, it is important for owners of patent applications relating to AI systems to make appropriate assessments of the level of risk of their systems, paying particular attention to cases where user interfaces are provided through which personal information can be entered or where tools (e.g. microphones or cameras) are provided programmed to automatically acquire personal information continuously over time.
In such cases, it will indeed be of paramount importance to assess how such information is used by the particular AI system, especially when the AI system makes decisions or predictions on the basis of such information, in order to avoid, for instance, potential unfair or discriminatory practices against particular categories of people or entire populations. Similarly, all those cases in which AI systems create user profiles or collect user preferences should also be assessed, as this could potentially fall within the critical scope of using such data to make assessments on various aspects of people's lives.
Besides the cited obligations, there remains an effort for legislators to promote the responsible development of artificial intelligence and to protect intellectual property rights, with the enforcement authorities called upon to protect the legitimate interests of the holders of such rights, be they patents or trade secrets.
For instance, the Regulation provides for free access for SMEs to so-called regulatory test spaces (sandboxes), which are to be set up by member state authorities in order to provide an environment for the development, training, testing and validation of AI systems under the supervision and support of national authorities. Regulatory testing spaces should also allow testing of AI systems under real-life conditions.
The Regulation applies to all entities offering products using artificial intelligence aimed at the European market. This also includes non-European players, thereby ensuring a level playing field that prevents non-EU companies from gaining a competitive advantage due to less stringent standards applied outside the EU itself.
It is clear from the above that the aspects of the AI Act and the issues posed by it are numerous and multifold.
Applicants should therefore be made sensitive to the importance of seeking expert advice in relation to their AI patent applications, software and trade secrets, and our team at Interpatent would be glad to assist them in this respect.
Back to News
These fines range from 35 million euros (or 7 per cent of turnover) for the most serious violations, relating to systems with unacceptable risk, to 15 million euros (or 3 per cent of turnover) for violations relating to data security and management, e.g. failure to provide documentation and information to the authorities, to 7.5 million euros (or 1 per cent of turnover) for providing inaccurate or misleading information.
In the light of this Regulation, it is important for owners of patent applications relating to AI systems to make appropriate assessments of the level of risk of their systems, paying particular attention to cases where user interfaces are provided through which personal information can be entered or where tools (e.g. microphones or cameras) are provided programmed to automatically acquire personal information continuously over time.
In such cases, it will indeed be of paramount importance to assess how such information is used by the particular AI system, especially when the AI system makes decisions or predictions on the basis of such information, in order to avoid, for instance, potential unfair or discriminatory practices against particular categories of people or entire populations. Similarly, all those cases in which AI systems create user profiles or collect user preferences should also be assessed, as this could potentially fall within the critical scope of using such data to make assessments on various aspects of people's lives.
Besides the cited obligations, there remains an effort for legislators to promote the responsible development of artificial intelligence and to protect intellectual property rights, with the enforcement authorities called upon to protect the legitimate interests of the holders of such rights, be they patents or trade secrets.
For instance, the Regulation provides for free access for SMEs to so-called regulatory test spaces (sandboxes), which are to be set up by member state authorities in order to provide an environment for the development, training, testing and validation of AI systems under the supervision and support of national authorities. Regulatory testing spaces should also allow testing of AI systems under real-life conditions.
The Regulation applies to all entities offering products using artificial intelligence aimed at the European market. This also includes non-European players, thereby ensuring a level playing field that prevents non-EU companies from gaining a competitive advantage due to less stringent standards applied outside the EU itself.
It is clear from the above that the aspects of the AI Act and the issues posed by it are numerous and multifold.
Applicants should therefore be made sensitive to the importance of seeking expert advice in relation to their AI patent applications, software and trade secrets, and our team at Interpatent would be glad to assist them in this respect.
Back to News