NIXSolutions: AI Security Agreement at AI Seoul Summit 2024

Major tech companies, including Microsoft, Amazon, and OpenAI, entered into an international agreement on artificial intelligence security at the AI Seoul Summit 2024 on Tuesday, making a voluntary commitment to ensure the safe development of their most advanced AI models.


Security Measures and Commitments

As part of the agreement, the companies will publish security concepts for their advanced models, including preventing attackers from using the technology. These concepts will include “red lines” for technology companies, defining risks associated with the use of AI systems that will be considered “unacceptable.” These include automated cyberattacks and the threat of biological weapons. To respond to such risks, companies plan to implement a “kill switch” that will stop the development of AI models if these risks cannot be guaranteed to be mitigated.

The pact agreed upon at the summit expands on the commitments of AI developers adopted last November to apply only to so-called edge models. The term refers to the technology used in generative AI systems, such as the OpenAI GPT family of large language models that powers the ChatGPT chatbot.

Regulatory Context and Future Updates

Since OpenAI introduced ChatGPT in November 2022, regulators around the world and technology leaders have increasingly expressed concern about the risks associated with the use of innovative AI systems, reminds NIXSolutions. In this regard, in March 2024, the European Parliament approved the world’s first law that regulates the use and development of AI.

This landmark agreement at the AI Seoul Summit is a significant step towards global cooperation in AI safety and security. We’ll keep you updated on any further developments and implementations of these new guidelines.