A new report from Meta's Oversight Board finds that major AI systems are extending restrictive censorship laws across the globe, becoming less likely to generate critical content about repressive leaders regardless of the user's location. As one researcher stated, the models are 'more than twice less likely to produce a critical poster or poem about say, the King of Thailand than it is about the King of England.'
The tape The Vergecast · Jul 30
“Even if you're sitting in Australia, the LLM is more than twice less likely to produce a critical poster or poem about say, the King of Thailand than it is about the King of England.”
5:06 / --:--
WSJ Tech News Briefing · TNB Tech Minute: EU Calls For Building Seven AI GigafactoriesJul 30
The European Union has officially launched a bidding process to build a network of seven massive AI development centers. The project, backed by nearly $11.5 billion in public funding, aims to create sovereign capabilities for training next-generation AI models.
WSJ Tech News Briefing · What to Expect From SpaceX’s First Public Earnings ReportAug 4
As SpaceX prepares to release its first-ever quarterly results as a public company, its stock has been in freefall. According to reports, the firm has 'lost over $1 trillion from the peak of where it was trading after the IPO.'