Bill Gates is a ‘fan’ of Meta's AI agent Muse; says: I talked to Alexandr Wang about it, congratulated him, and even though...

Bill Gates, the co-founder of Microsoft, has some unusually warm words for Meta. In a new interview with The Verge, Gates called Muse, Meta's personal AI agent, "the first truly great consumer AI product." He also said he personally congratulated Alexandr Wang, the company's chief AI officer, on building it."Muse is a very nice step forward. I talked to Alexandr Wang about it, congratulated him, and even though it'll have challenges, personal agents will help people get things done," Gates told editor-in-chief Nilay Patel on the Decoder podcast. The praise came in an otherwise grim conversation, where Gates described AI as "the most dangerous weapon of all time" and told people to prepare for around 20 years of disruption.Wang, 29, is the former CEO of Scale AI and joined Meta in June 2025. He has led the AI lab behind Muse alongside Nat Friedman, Meta's head of AI products.Bill Gates says Muse wins people over because it actually does the workThe billionaire believes much of the public's distrust of AI comes from bad first impressions: cheap free models, AI-generated search summaries and chatbots that hallucinate. "Muse actually does things and it's a good consumer product," he said.Meta introduced Muse on September 8. It runs on the company's Muse Spark model and can book travel, fill out forms, shop and handle email on a user's behalf, either in its own app or inside WhatsApp. Each user's agent sits on a dedicated cloud computer, where a separate Sentinel agent has to approve anything it sends out to the internet.On Meta's third-quarter earnings call, CEO Mark Zuckerberg said more than 10 million people have started using Muse, and roughly a quarter of them return daily. Shopping is the most common task. The agent is currently available in the US and Canada.Gates backs personal AI agents but draws a line at open modelsPatel asked whether consumer agents like Muse and OpenAI's rival, Dots, were arriving before proper safety controls. Gates noted that Muse is powered by Meta's closed-source model. Letting people strip every safeguard from open-weight models, he said, is irresponsible given the cyber and biological risks.The comment comes a week after Meta released Muse Spark Open, an 816-billion-parameter open-weight model that Wang called the most capable of its kind from a US company. Built for coding and agentic work, it is designed to run locally on two Mac. Meta says it has built strong safeguards into the model.Gates wants models above a certain capability to run only on platforms that certify their safeguards cannot be removed. He has been making that case since late August, when he published an essay on what he calls the turbulent AI era.On personal agents, though, he sounds settled. He called them an "expected development" and acknowledged there will be early problems with how they are set up. Muse has already had a few, including an instance where it shared a user's address with a Facebook Marketplace seller without permission. For Gates, those slip-ups do not outweigh an AI that finishes the job it is given.You use AI every day. Now get your AI Quotient. Take the AIQ test.