Multibot, the AI ​​agent that “actually does things,” is the new technology obsession

An open source AI agent that “actually does things” is on the way out, with people across the web sharing how they use the agent to do a whole host of things, e.g. Manage remindersregister Health and fitness dataAnd even Communication with customers. tool, It’s called a multipot (formerly known as Clawdbot), runs natively on a variety of devices, and you can ask it to perform tasks for you by chatting with it via WhatsApp, Telegram, Signal, Discord, and iMessage.

Federico Vitici Mac Stories He highlighted how he installed Moltbot on his M4 Mac Mini and turned it into a tool that provided daily audio summaries based on his activity in the Calendar, Notion, and Todoist apps. Someone else I prompted Moltbot to give itself an animated face, and it added a sleep animation without prompting.

Moltbot routes your request through the AI ​​provider of your choice, such as OpenAI, Anthropic, or Google. Like many of the AI ​​agents we’ve seen so far, Moltbot can fill out forms within your browser, send you emails, and manage your calendar — but it does so more efficiently, at least According to some of people using the tool.

There are some caveats, though; You can also give Motlbot permission to access your entire computer system, allowing it to read and write files, run shell commands, and execute scripts. The combination of admin-level access to your device and your app credentials can pose significant security risks if you’re not careful.

“If your autonomous AI agent (like MoltBot) has administrative access to your computer and I can interact with it by sending you a direct message on social media, well, now I can try to hijack your computer in a simple direct message,” Rachel Toback, CEO of SocialProof Security, says in an email to Edge. “When we grant admin access to autonomous AI agents, they can be compromised through spot injection, a well-documented vulnerability that has not yet been resolved.” Instantaneous injection attack This happens when a bad actor manipulates the AI ​​using malicious claims, which they can make directly to the chatbot or embed within a file, email, or web page that is fed to a large language model.

Find out about Jamison O’Reilly, security professional and founder of cybersecurity company Dvuln Those private messagesThe account credentials and API keys associated with Moltbot were left exposed on the web, which could allow hackers to steal this information or exploit it to launch other attacks. O’Reilly says he reported the issue to Moltbot’s developers, who have since released a fix. according to Record.

One of Moltbot’s developers said on X’s website that the AI ​​agent is “powerful software with a lot of sharp edges.” Warning to users You should “read the security documents carefully before operating them anywhere near the public Internet.”