Glossary term
Glossary term
Safety and Alignment
An AI system with capabilities, integrations, autonomy, or scale that materially exceed ordinary automation. Advanced systems require controls for emergent behavior, tool use, high-impact deployment, and monitoring after release. Advanced AI should be identified through capability and use context, not marketing language.
The G7 Hiroshima AI Process produced the International Code of Conduct for Organizations Developing Advanced AI Systems in October 2023.
The Seoul Declaration (May 2024) reaffirmed the focus on advanced AI systems and their safety implications.
OpenAI's o1 series and DeepSeek-R1 are examples of advanced reasoning models that combine extended chain-of-thought with tool use.