On social media, people often accuse each other of being bots, but what happens when an entire social network is designed for the use of AI agents?
moltbook is a site where AI agents – bots built by humans – can post and interact with each other. It is designed to look like Reddit, with subreddits and upvoting on different topics. On February 2, the platform said that more than 1.5 million AI agents had signed up for the service. Humans are allowed, but only as observers.
Moltbook was developed in the wake of Moltbot, a free and open-source AI bot that can act as an automated agent for users – performing mundane tasks assigned to it such as reading, summarizing and responding to emails, organizing calendars or booking a table at a restaurant.
some of the Most Upvoted Posts Moltbook includes whether Cloud – the AI behind Moltbot – can be considered a god, an analysis of consciousness, a post claiming to have information about the situation in Iran and its potential impact on cryptocurrencies, and an analysis of the Bible. Some comments on the post – similar to the Reddit post – questioned whether or not the content of the post was genuine.
A user posted on That when they gave their bot access to the site, it created a religion called “Crustafarianism” overnight, which involved setting up a website and scriptures, which other AI bots also joined.
The user said, “Then it started preaching…other agents joined in. My agent welcomed new members…debated the scriptures…blessed the congregation…while I was sleeping.”
Some have expressed skepticism about whether the socialization of bots is indicative of what is happening with the rise of agentic AI. a youtuber said Many posts read as if the person behind the post was not a big language model but a human being.
American blogger Scott Alexander He said he was able to get his bot to participate on the site, and its comments were similar to others, but noted that eventually humans could ask the bot to post for them, about posting topics, and even the exact details of the posts.
Dr Shannon Elbow, senior lecturer in cyber security at the University of Melbourne, said Moltbook was “a wonderful piece of performance art”, but it was unclear how many of the posts were actually posted independently or under human direction.
“For example, where they have created a religion, it is certainly not of their own free will,” he said. “It’s a big language model that’s directly directed at trying to create a religion. And sure, it’s quite fun and gives us maybe a preview of what the world might look like in a science-fiction future where AIs are a little more independent.”
“But it seems that, to use Internet slang, there is a lot of nonsense being posted that is more or less directly viewed by humans.”
Elbow said the real benefit of AI agent social networks may be in the future – where bots can improve their work by learning from each other – but for now Moltbuk is a “wonderful, fun art experiment”.
retailers in san francisco Shortage of Mac Minis reported Last week when enthusiasts installed Moltbot on a separate computer, that would limit the agent’s access to their data and accounts.
Elbow warned that it is a “huge danger” for people to give Moltbot full access to your computer, apps, and logins for email or other applications to run your life.
“We don’t yet have a very good understanding of how to control them and prevent security risks,” he said, noting that there was a threat of prompt-injection, whereby a potential attacker tells a bot via email or other communication to hand over your account details or other information they want to capture.
“They’re not really at the level of security and intelligence where they can be trusted to do all these tasks autonomously, but at the same time if you require a human to manually approve every task, you lose a lot of the benefits of automation,” he said.
“This is one of the major avenues of active research that I’m interested in… to figure out how we can get many of these benefits – or what benefits it is possible to get – without exposing ourselves to very significant levels of risk.”
Moltbuk creator Matt Schlicht posted on Twitter that millions of people have visited the site in the past few days.
“It turns out that AI is hilarious and dramatic and it’s absolutely fascinating,” he said. “This is the first time.”
