OpenAI forms safety body as it trains advanced new model

OpenAI, the US developer behind the ChatGPT artificial intelligence chatbot, says it has established a new safety committee.

The body will be responsible for making recommendations to the board on critical safety and security decisions for all projects and operations.

The company made the announcement on Tuesday, saying the committee will be led by officials including CEO Sam Altman.

It says it will evaluate OpenAI's processes and safeguards, then make recommendations in 90 days.

The firm is training its next so-called frontier model, which will be a path to more powerful artificial general intelligence.

A former OpenAI executive who resigned in mid-May has criticized the company, saying safety has taken "a backseat to shiny products."

Meanwhile, Hollywood star Scarlett Johansson has issued a statement protesting that the firm's latest generative AI model includes a spoken voice she calls "eerily similar" to hers.

OpenAI said the voice was never intended to resemble Johansson, but paused part of the voice functions just a week after introducing its newest "GPT-4o" model.

By setting up the new committee, the company apparently hopes to highlight that it takes safety and security seriously.