Researchers Advised to Help Address Gaps in EU's AI Act Before Finalization
• There are gaps in the EU's proposed AI Act regarding risk classification and self-assessment by AI developers that need addressing.
• Researchers can help shape the regulations by providing evidence on AI risks and advising on independent verification of high-risk AI.
• The new EU AI Office should establish a third-party system to verify high-risk AI data/models.
• Maximum transparency is essential in AI regulations, as is regulatory independence from AI developers.
• Researchers have a short window to provide expertise and advise the EU before the AI Act is finalized in around 2 years.