Weeks before the Biden administration comes to a close and President-elect Donald Trump takes the reigns, the government plans to issue another set of regulations restricting AI chip exports, according to Bloomberg. As usual, the move is to maintain control over AI development and protect national security and the US' allied nations. However, the rules have drawn backlash from tech companies like Nvidia and US nation allies like Israel, citing economic concerns and potentially being a double-edged sword, as some fear it could drive nations toward Chinese alternatives.
Silicon Valley AI startup Anthropic is, according to The Wall Street Journal, reportedly nearing a deal to raise $2 billion, boosting its valuation to $60 billion. Led by Lightspeed Venture Partners, the fundraiser would mean that Anthropic total raised would be around $15.7 billion. Valued previously at $18 billion last year, Anthropic has been one of OpenAI's most prominent mainstream competitors, with backers including the likes of Alphabet and Amazon and revenue sitting at $875 million annually.
Following an unassuming note all the way back in April of 2024, xAI, Elon Musk's AI startup behind the Grok, recently updated its FAQ to detail "Unhinged Mode." The feature is designed to deliver intentionally "objectionable, inappropriate, and offensive" responses akin to an "amateur stand-up comic," reads the page. Though the mode isn't live, it lines up with Musk's original vision of Grok as an edgy, unfiltered AI model. However, it has been a long road to get here, as Musk has criticized Grok's training data for leaning politically left and pledged to steer it toward neutrality, calling the current iteration a beta version.
In the copyright lawsuit Kadrey v. Meta, plaintiffs allege that Mark Zuckerberg approved using the pirated dataset LibGen to train Meta's Llama AI models despite teamwide warnings about massive legal concerns. Unsealed court documents claim Meta engineers stripped copyright markers and torrent-shared LibGen content to obscure infringement. Though Meta has invoked fair use as its defense, plaintiffs argue the company knowingly bypassed legal data acquisition methods. Still, for the time being, the court case is far from settled, and the case only reflects Meta's older Llama models, not its recent releases.
As reported first by Bloomberg News, the Dutch government is in talks with GPU hardware giants Nvidia and AMD to establish an AI facility featuring a supercomputer for AI development and application. Dutch Economic Affairs Minister Dirk Beljaarts confirmed a deal with Nvidia for hardware and technical know-how, saying that "this deal brings building a Dutch AI facility a lot closer." Despite Nvidia signing a deal, AMD has not yet responded to any comments or questions regarding its position.
Just recently, The White House introduced the Cyber Trust Mark (CTM), a shield-like label indicating cybersecurity compliance for internet-connected devices anywhere from smart thermostats to baby monitors. Products with the label must meet certain cybersecurity standards set by the US National Institute of Standards and Technology, with testing conducted by approved labs. The move is to help consumers pick out safer devices, though CTM participation is voluntary. A government plan mandates purchasing only CTM products starting in 2027, with labeled items expected in stores later this year.
AWS recently announced plans to invest at least $11 billion in Georgia to further expand its cloud computing and AI infrastructure, which is estimated to create 550 jobs. Due to low electricity costs, tax incentives, and existing fiber-optic infrastructure, Georgia has become an attractive hotspot for data center construction for other tech giants like Google and Microsoft, too. However, locals worry about real estate competition and the environmental impact of such moves. The AI boom continues to fuel data center demand, with AI projected to account for 19% of power use in such facilities by 2028.