Many of the top AI firms are joining a new U.S. consortium focused on safety considerations for generative AI.
Members of the U.S. AI Safety Institute Consortium include OpenAI, Alphabet’s Google, Anthropic and Microsoft, Meta Platforms, Amazon.com, Apple and Nvidia.
The consortium will develop guidelines for risk management, capability evaluations, safety and security, watermarking of AI-generated content, and “red-teaming” cybersecurity exercises to simulate threats and identify new risks. The move comes after an October executive order related to AI and a December announcement from the Commerce Department that it was moving to write standards and guidance for safe AI testing and use.
Major AI companies already said last year they would begin watermarking AI-generated content to promote safety.