Meta's new AI agent, Muse, is designed to perform tasks on behalf of users, but experts are warning that the rapid growth of AI agents could create new problems for websites, businesses and internet users.
Meta introduced Muse as a personal AI agent capable of carrying out tasks such as sending emails, booking travel and interacting with online services. The company says users remain in control of which applications Muse can access and that the system seeks permission before sensitive actions such as purchases.
The technology represents a shift from traditional chatbots. Instead of simply answering questions, Muse can use a browser and take actions on a user's behalf. Meta says it runs inside a dedicated virtual machine designed to separate the agent from a user's personal data and provide additional security controls.
However, technology and privacy experts have raised concerns about what could happen if millions of AI agents begin interacting with the internet at the same time. One concern is that automated systems could overwhelm websites by making large numbers of requests, bookings or enquiries much faster than humans normally would.
Ticketing and appointment systems could be particularly affected. Experts cited in recent reporting have warned that people could use AI agents to compete for limited concert tickets, restaurant reservations and government appointments, potentially making existing problems involving automated bots even more difficult to manage.
Small businesses could also face challenges if automated agents begin sending large volumes of enquiries or requests. Websites built around advertising and human visitors may have to adapt if AI agents increasingly access information without viewing advertisements or generating conventional web traffic.
Security is another concern. Meta says Muse has undergone extensive safety testing and includes protections designed to prevent the agent from accessing passwords and payment information directly. The company also says users can review what the agent has done and disconnect connected services.
Security researcher Patrick Wardle, however, said he discovered a vulnerability in Muse shortly after its release that could potentially have allowed an attacker to take control of the application and access connected devices. Meta reportedly fixed the flaw and said the product had been subjected to months of additional privacy and security testing.
Privacy has also become part of the debate. Meta says users can opt out of having their interactions with Muse used to train its AI models, while the company says personal data used for training is processed to remove identifying information.
Despite the concerns, Muse demonstrates why technology companies are investing heavily in AI agents. The systems could save users time by handling routine digital tasks, but their wider adoption may require websites and online services to change how they manage automated activity.
The debate is therefore moving beyond whether AI can answer questions. As agents such as Muse gain the ability to act independently online, businesses, regulators and technology companies will increasingly have to consider how the internet should work when humans are no longer the only users interacting with it.