Artificial intelligence (AI) capabilities are accelerating and its applications are expanding in business, the economy, and cybersecurity, but the regulatory frameworks governing this technology still raise questions about its ability to keep pace with emerging risks. In this context, Microsoft President Brad Smith called for stricter global rules to regulate AI, drawing a striking comparison between the safety requirements of this technology and the regulations imposed on everyday products like milk and washing machines.
Hot Topics
The human body ages in leaps and bounds Scientists identify two critical ages
From your phone screen to your data How do malicious apps sneak in and steal your secrets?
From product safety to the risks of smart systems
Smith’s comparison is based on the difference in the nature of control between traditional products and digital technologies. In the food and household appliance sectors, there are rules relating to safety and quality, procedures for inspection and monitoring, and legal responsibilities when damage occurs, although the details differ from one country to another.
As for artificial intelligence, its capabilities change rapidly, and its risks vary according to the model, how it is used, and the permissions granted to it. A chat system may produce misleading information, reveal sensitive data, or an intelligent agent may perform unintended digital actions if it is granted access to files, accounts, or work tools.
The comparison does not imply that artificial intelligence is subject to a complete lack of regulation, as laws and governance frameworks already exist in a number of countries, but it highlights the challenge of translating general safety principles into verifiable and accountable tests and procedures that are commensurate with the speed of systems development and the breadth of their uses.
These tests are gaining importance as artificial intelligence moves from answering questions to performing tasks on behalf of users, such as handling files, software, and digital systems. In these cases, the error may not be limited to providing incorrect information, but may extend to performing an unwanted action or accessing data that the system should not have access to.
Tags:
technology