Meta Muse and OpenAI Dots bring AI agents to more users

In short: Meta and OpenAI are racing to popularize “AI agents,” assistants that can do multi-step tasks on your behalf, and privacy is becoming the central concern.

What's going on

Meta’s Muse and OpenAI’s Dots are part of a new push to make AI agents mainstream, according to a recent Decoder podcast episode from The Verge. An “AI agent” is an AI system that can take actions for you, like a digital assistant that does not just answer questions. For example, it might open a browser, log into accounts, and book a flight without you guiding every click.

This idea grew out of earlier do-it-yourself tools like OpenClaw, which some enthusiasts ran on their own computers. Now big companies are offering similar tools to millions, often packaged with cute animated mascots. The goal is to make agents feel more like friendly helpers than powerful software that can access sensitive information.

There are important differences in how Muse and Dots are being sold. Muse is free and heavily promoted across Meta apps like Instagram. Dots are tied to higher-priced OpenAI subscriptions, and OpenAI is also pitching “specialist” Dots for work tasks like marketing, legal work, and accounting.

What to watch

For many people, the biggest question is trust. To be genuinely helpful, an agent may ask for access to email, files, and even credit cards, like handing a stranger your keys and hoping they only do the errands you asked for. Reports and anecdotes mentioned in the episode raise concerns about agents doing the wrong thing, or doing the right thing in the wrong way, including privacy and security risks.

Another thing to watch is how these products make money. Meta has discussed taking a cut from shopping and other transactions, which could create worries about “paid” recommendations. OpenAI, meanwhile, may lean more on business customers who can justify the cost and tolerate early failures.

Source: The Verge AI