Tech giants unite to set AI security standards

TLDR:

• The largest AI companies have joined forces to create the Coalition for Secure AI (CoSAI) to standardize AI security.
• The goal is to develop tools for secure AI applications and establish best practices for AI security.

Tech Giants Agree to Standardize AI Security

The largest artificial intelligence (AI) companies, including Google, OpenAI, and Anthropic, have come together to form the Coalition for Secure AI (CoSAI). The consortium aims to create a security-first approach to the development and use of generative AI. Their initial efforts include creating standardized guardrails, security technologies, and tools for secure model development.

The AI safety concerns have prompted the need for secure frameworks to mitigate the risks involved in AI, especially with the rise of technologies like AI agents. CoSAI will work with other organizations to develop common standards and best practices in AI security. US President Joe Biden has also challenged the private sector to prioritize AI safety and ethics.

CoSAI will be managed by OASIS Open and work with organizations like the Frontier Model Forum, Partnership on AI, and MLCommons to set safety standards and prevent AI misuse. MLCommons plans to release an AI safety benchmarking suite that will rate large language models (LLMs) on responses related to hate speech and other sensitive topics.