According to Bloomberg News on the 18th, the European Union is planning to adopt a comprehensive three-level system to supervise generative artificial intelligence models and systems, and adopt different ways to supervise and test different levels of models. This marks a key step for Europe to cope with the rapidly developing technology, which will make the EU the first western government to implement mandatory rules for artificial intelligence.
The proposed three-level supervision system includes the following categories:
1. All basic models
2. “Very powerful” basic model
3. Large-scale general artificial intelligence system
In addition, the European Union’s “Artificial Intelligence Act” is also being formulated. If approved, the bill will require systems with the ability to predict crimes or screen job applications to accept risk assessment and comply with specific regulations. Relevant negotiators intend to improve the legislation at the meeting on October 25th and finalize it before the end of the year. The EU said that more discussions are needed to establish safeguard measures to prevent the generation of illegal and harmful content. The newly established artificial intelligence office will supervise the compliance with the rules, have the right to request documents, organize compliance tests, and suspend the model as a last resort.
Post time: Oct-20-2023