Here are the five most notable news stories of the week about artificial intelligence.

This week marked the start of the Halloween season, and in the realm of artificial intelligence, several significant events took place. Notable among them are OpenAI raising $6.6 billion, Nvidia launching an open-source language model, Google's Gemini Live expanding to multiple languages, the veto of an AI safety bill in California, and Meta modifying its smart glasses to invade privacy.

OpenAI has successfully secured an investment of $6.6 billion in its latest funding round, with contributions from existing investors like Microsoft and Khosla Ventures, as well as new participants such as SoftBank and Nvidia. This funding has raised the company's valuation to an impressive $157 billion, placing it among the world's most valuable private companies. If a proposed restructuring plan for profit is realized, CEO Sam Altman could gain over $150 billion in stock, making him one of the ten richest people globally. Additionally, OpenAI has launched Canvas, a collaborative feature similar to Anthropic's Artifacts.

On its part, Nvidia has introduced the LVNM 1.0 language model, an open-source software excelled in various vision and language tasks. The model, which boasts 72 billion parameters, aims to establish itself as a platform for other developers to create their own chatbot applications, rather than directly competing with other leading models.

In another development, Google has expanded its Gemini Live assistant to allow conversations in nearly forty languages, including French, German, Portuguese, Hindi, and Spanish. This trend aligns with Microsoft's adoption of similar features with its Copilot tool and other voice interaction applications from ChatGPT and Meta, providing users with a more interactive and natural experience.

Regarding regulation, California Governor Gavin Newsom decided to veto bill SB 1047, which aimed to establish safety regulations for AI models. In his letter to legislators, Newsom argued that the legislation focused too much on large language models, overlooking the possibility that smaller, specialized models could pose equivalent or greater risks.

Finally, a couple of computer science students from Harvard managed to modify Meta's smart glasses to identify and search for information about anyone appearing in their field of vision. This experiment, called I-XRAY, uses facial recognition software to obtain personal data about individuals, raising serious privacy implications. Although the students have no plans to make the source code public, their demonstration suggests an imminent risk that others could replicate this system.