As AI capabilities grow, so does the effort to govern them.
Company Safety Frameworks
Leading developers have published frameworks setting capability thresholds that trigger additional safeguards, evaluations and security measures before training or deployment continues.
National Regulation
Governments are taking different approaches: risk-based rules for AI applications, transparency and reporting requirements for frontier models, and sector-specific rules in areas such as finance and health.
AI Safety and Security Institutes
Several governments have established bodies to evaluate frontier models and research safety.
International Cooperation
International summits, declarations and scientific reports aim to build shared understanding of risks. Proposals for deeper cooperation include shared evaluation standards and compute monitoring.
Key Debates
- How to regulate without stifling beneficial innovation.
- Open versus closed model releases.
- Whether and how to monitor large training runs.
- Liability for AI harms.
Staying Informed
The landscape changes quickly. For compliance, consult current regulations in your jurisdiction.