Gates praises Meta Muse while urging stronger safeguards for powerful AI

His Decoder interview pairs enthusiasm for personal agents with concern about removing safeguards from capable models.

Bill Gates praised Meta’s Muse in an October 22 interview with Decoder, describing personal agents as a useful advance while arguing that more capable AI demands stronger safety institutions. His endorsement of a consumer product sits alongside a broader warning about cyber and biological misuse when safeguards can be removed.

The distinction matters. Gates was discussing both everyday assistants and the risks of powerful models. His remarks do not establish that Muse caused a particular safety incident, or that personal assistants should be prohibited.

From answering questions to taking action

Meta describes Muse as an agent that can perform tasks through a dedicated virtual computer. Its announcement describes browsing, completing forms and helping with longer-running goals, with approval checks for sensitive actions such as purchases or outgoing email. Those are the company’s descriptions of its design, rather than an independent assessment of every real-world interaction.

An assistant that acts creates a different relationship with its user. A mistaken answer can be checked before it is used; a mistaken action may already have changed a booking, shared information or contacted another person. Useful delegation therefore depends on understandable permissions and clear moments when a person makes the decision.

Enthusiasm and oversight can coexist

Gates’ argument leaves room for useful products while calling for a stronger response to the technology’s wider effects. The consumer test is practical: can someone understand what the assistant is authorised to do, inspect what happened and change its access when their needs change?

For organisations evaluating personal agents, that suggests starting with a bounded task and judging the complete handover between person and software. Convenience is valuable when the person can also recognise the consequences of granting access. A product endorsement and a call for oversight can support the same goal: making capable assistants useful enough to trust with carefully chosen work.

Sources

- [Bill Gates says to prepare for 20 years of tumultuous AI disruption](https://www.theverge.com/podcast/1015408/bill-gates-ai-interview-microsoft-openai-anthropic-risk-policy-regulation) — The Verge / Decoder - [Introducing Muse: The World’s First Personal AI Agent Built for Everyone](https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/) — Meta

*Image: AI-generated editorial illustration; not a photograph of a documented event or an authentic product screen.*