Heise tests OpenClaw: "The World's Most Dangerous Software?"
Germany's Heise Group has tested an AI assistant named OpenClaw, which developer Peter Steinberger describes as potentially "the world's most dangerous software." The tool has garnered significant hype and attention.
Heise Gruppe GmbH & Co. KG's publication heise online has reviewed OpenClaw, also known as Moltbot and ClawdBot, an AI assistant constructed with coding agents. The project has generated substantial interest, evidenced by over 118,000 stars on its GitHub repository.
Heise's technology desk explored the tool and interviewed its creator, Vienna-based Peter Steinberger. Steinberger indicated that the software enables the automation of various daily tasks, including responding to emails and sending messages across platforms like Telegram and Discord. He controversially labeled OpenClaw as potentially "the world's most dangerous software," highlighting its significant capabilities and inherent risks.
The article details the installation process of OpenClaw on a Fedora Linux system, noting the convenience of a one-line command setup. However, this ease of installation raises security concerns due to its lack of transparency. The software is rapidly evolving, aiming to integrate multiple applications via APIs. Steinberger himself has expressed reservations about the project's unfinished state and its potential for unpredictable behavior.
Heise's report seeks to understand the reasons behind the intense interest in OpenClaw and to critically assess its implications. While the tool offers considerable automation potential, its developmental stage and associated risks, including Steinberger's allusion to 'Skynet,' remain key points of examination.