OpenClaw, the new darling of AI agents, is already in the sights of the giants.

Behind its somewhat unusual name, OpenClaw has made a spectacular entrance into the tech world. The open-source project, born less than six months ago from the hands of an Austrian developer, presents itself as a platform dedicated to artificial intelligence "agents"—software that no longer just chatters but actually acts. And when Jensen Huang, CEO of Nvidia, declared it the "next ChatGPT," the spotlight immediately shone on it, with all the expectations, fantasies, and hype that entails.

In practical terms, these agents are presented as being able to use a computer "like a human." Booking a ticket, ordering a vehicle, preparing purchases, tackling administrative tasks, or even helping to design everyday projects, such as a home improvement plan: the idea is simple, almost appealing. A written instruction, sent via a messaging app like WhatsApp, and the agent carries out the operations for you, provided they have the necessary access to the relevant services. Fast. Convenient. A little confusing too, let's be honest.

Agents who click for you, a promise that attracts as much as it worries

This shift towards "executor" AI is far from insignificant. Since the widespread adoption of chatbots in late 2022, the race has shifted: writing an email is fine, but managing an entire process across multiple applications is considerably more cost-effective… and more sensitive. Large tech companies are therefore closely monitoring these solutions, as they impact everything from customer support to work organization and the management of personal tasks—those things that fill our days without us even realizing it.

The sticking point remains, the one we sometimes avoid because the enthusiasm is so contagious: security. An agent using your credentials can also, by default, access sensitive services, including payment accounts, and the slightest vulnerability can be costly, both financially and in terms of trust. In Europe, these practices are growing under the watchful eye of regulators, between the GDPR and the AI ​​Act, with transparency and risk management requirements that are far from a technical detail. OpenClaw may be a promising technology, but it could also become a large-scale test: how far are we willing to delegate our clicks, and therefore a part of our digital lives?