Today in AI: Former OpenAI director calls for AI auditing; Mistral AI seeks $5 billion valuation; and more

Our daily TLDR of important AI news you should know about.

The world of artificial intelligence is constantly evolving, with new developments, challenges, and investments emerging. Here's a roundup of today's developments in the field of AI.

Helen Toner, a former board member of AI research lab OpenAI, believes companies developing AI should be required to disclose more information about their projects. In a recent speech at the TED conference in Vancouver, Toner emphasised the need for AI companies to be open about their technology's capabilities and potential risks. She proposes "AI auditors" to examine these systems, ensuring companies are held accountable and not left to self-assess.

French AI company Mistral AI is said to be in talks with investors for a new funding round. The company is aiming for a valuation of $5 billion and could raise hundreds of millions of dollars. Although specific investors haven't been named, Mistral already has backing from Andreessen Horowitz, Lightspeed Ventures, and even Microsoft (who also invests in Mistral's competitor, OpenAI).

Chipmaker Intel is releasing two new AI chips, HL-328 and HL-388, specifically designed for the Chinese market. These chips are based on Intel's recently launched Gaudi 3 line and boast similar features like on-chip memory, but will have noticeably slower performance to meet export regulations. The HL-328 is expected in June, with the HL-388 following in September.

Google's AI chief, Demis Hassabis, hinted at a massive investment in AI exceeding $100 billion over time. This comes amidst reports of Microsoft planning a similar supercomputer project. Hassabis, however, remained tight-lipped about specific spending figures but claimed Google has a computational edge over competitors.

Meta's independent Oversight Board is investigating how the company handled two incidents of AI-generated pornographic photos of female celebrities on Facebook and Instagram. These cases will help them evaluate how effective Meta's policies are against deepfakes and other AI-powered pornography.