Meta and OpenAI push AI agents, and safety questions follow

Meta and OpenAI are each pushing an AI agent, and Bloomberg Technology has published an explainer on both. Meta's is called Muse. OpenAI's is called Dots. The explainer also covers what these agents can do and the safety worries that come with them.

What an agent is

An AI agent is software built to act like a digital assistant for you. The tech industry has long been obsessed with building a powerful assistant, whether a chatbot, a smart speaker or a robot. Bloomberg says artificial intelligence offers the promise of finally reaching that goal.

Who it affects

Anyone who may be offered an assistant that does things on their behalf. That includes everyday phone users, families and business owners who might hand tasks to software. The core question is how much control to give it.

What is not known yet

The full text of Bloomberg's explainer was not available for this report. That means the specific abilities of Muse and Dots, their release dates, their prices and the exact safety concerns are not confirmed here. No details on how either company plans to limit what its agent can do have been provided.

What to do

Before letting any AI agent act for you or your business, learn what it can do and what safety risks come with it. Questions worth asking:

• What can it do without asking me first?
• What personal or business information can it see?
• Can I review or undo what it does?