AI in the world
Posted: Sun Feb 09, 2025 8:04 am
New legislative initiatives and trends in the development of AI in information security
Yuri Silaev, Head of R&D Department, Infotactics Group | 02/09/2024
An active discussion about the need to regulate artificial intelligence has begun in the world. The first legislative initiatives in this area have already been published, adopted at the government level in the European Union, China and the United States. Let's consider the essence of such events, and also predict how AI will develop in information security.
The AI law under consideration in the European Union contains a risk-oriented approach to the gradation of AI systems. This vector allows for selective control of AI decisions in various south africa mobile database of the economy. The imposition of fines alone for violating the rules for compliance with the gradation of AI systems by risk levels is insufficient, because if we take large corporations, such fines will not be an obstacle for them to deliberately understate the risk of AI systems. The regulatory sandboxes described in the law also raise questions about the methods of interaction with regulators and feedback on the development and subsequent compliance with methodological recommendations for a comprehensive audit of AI systems. Perhaps such rules will be developed by each EU member state separately, but the lack of a wide base of methodological documents at the moment does not allow us to talk about the safe use of AI systems.
The US legislative initiative in terms of safe AI includes deep control over all decisions in the field of AI safety by the government and various national centers. Such penetration of government agencies allows for the introduction of various "holes" or "bookmarks" into almost any solution based on AI technologies. And the vector of international cooperation specified in the decree through the national resource for research in the field of AI will allow for the spread of the introduction of such "bookmarks" including to various international decisions and research.
Yuri Silaev, Head of R&D Department, Infotactics Group | 02/09/2024
An active discussion about the need to regulate artificial intelligence has begun in the world. The first legislative initiatives in this area have already been published, adopted at the government level in the European Union, China and the United States. Let's consider the essence of such events, and also predict how AI will develop in information security.
The AI law under consideration in the European Union contains a risk-oriented approach to the gradation of AI systems. This vector allows for selective control of AI decisions in various south africa mobile database of the economy. The imposition of fines alone for violating the rules for compliance with the gradation of AI systems by risk levels is insufficient, because if we take large corporations, such fines will not be an obstacle for them to deliberately understate the risk of AI systems. The regulatory sandboxes described in the law also raise questions about the methods of interaction with regulators and feedback on the development and subsequent compliance with methodological recommendations for a comprehensive audit of AI systems. Perhaps such rules will be developed by each EU member state separately, but the lack of a wide base of methodological documents at the moment does not allow us to talk about the safe use of AI systems.
The US legislative initiative in terms of safe AI includes deep control over all decisions in the field of AI safety by the government and various national centers. Such penetration of government agencies allows for the introduction of various "holes" or "bookmarks" into almost any solution based on AI technologies. And the vector of international cooperation specified in the decree through the national resource for research in the field of AI will allow for the spread of the introduction of such "bookmarks" including to various international decisions and research.