New AI Law Enacted: Human Control Ensured

The European Union officially introduced the world’s first AI Law ( ai Act ), which is designed to protect the “fundamental rights” of citizens of the 27th Blok countries , simultaneously encouraging investments and innovations in the rapidly developing industry.

AI ACT has been developed for several years and is a comprehensive AI management guide in Europe. The law will affect primarily large American IT companies, but many organizations that are not even related to the technological sphere will fall under its effect. For example, companies using AI in their operations will also obey the new rules.

restrictions are based on four risk levels, and the vast majority of AI systems fall into the category of minimal risk, such as content recommendations or spam filters.

For each level of risk, the EU Law on AI establishes its requirements:

  1. Unacceptable risk:
  • These applications are completely prohibited. Examples include systems of social rating of citizens, predicting the police and technologies for recognizing emotions at workplaces or in schools.
  • High risk:
    • Requirements:
      • adequate assessment and minimization of risks.
      • The use of high -quality training data to minimize bias.
      • Routine management of activity logs.
      • Mandatory provision of detailed documentation on models to the authorities To assess their compliance with the requirements.
      • examples include autonomous vehicles, medical devices, decision -making systems on loans, educational systems for evaluations and systems of remote biometric identification.
  • Limited risk:
    • Requirements:
      • transparency of the use of AI. For example, chat bots should notify users that they interact with AI.
      • Examples include chat bots and systems that use AI to evaluate employees in the process of hiring.
  • Minimum risk:
    • Requirements:
      • Basic safety requirements and responsibility, without the need for special measures.
      • Examples include applications using spam filters, content recommendations, video games with AI elements and programs to help in weather forecasting.

      Compliance with all the rules will be controlled by the AI ​​office, which will be based in Brussels. Openai promised to follow Ai ACT and work closely with the new AI office The measure of the introduction of the law.

    /Reports, release notes, official announcements.