Moltbuk claims to be a social network for AI bots. But humans are responsible for its rapid growth.

It is branded as the world’s first social network for AI bots. But just a week after its launch, Moltbuk is being hotly debated in the tech world – and to what extent humans are responsible for the viral website’s rapid rise.

Launched in late January by tech executive Matt Schlich, Moltbuk’s user base (1.6 million strong, according to the site) claims to be composed entirely of AI agents – autonomous bots created to perform mundane digital tasks like writing emails or booking flights.

Many security researchers and journalists has already been proven that they can sign up for an account themselves, or Create an unlimited number of AI agents To join the site. Once they’re registered, agents post on Reddit-style forums as if they were just like real people.

Reaction to Moltbuk has ranged from reverence to horror, as people like tech billionaire Elon Musk have praised the site as a sign that artificial intelligence is beginning to overtake human cognitive ability. But some experts have deep doubts about that premise.

Mike Pepy, a New York-based technology critic and author, said, “I think Moltbuk is basically the latest in a long series of people becoming aware of artificial intelligence.” Against Platforms: Saving Digital UtopiaDuring an interview with CBC News.

What exactly is Moltbuk?

Schlich initially envisioned Moltbook as an experiment: a playground for AI agents that runs on OpenCloud, an open-source software that gives bots access and control over a user’s computer, allowing it to connect to apps like WhatsApp and Telegram.

When the bots hit Moltbuk’s Reddit-style forum, “something wonderful and strange happens,” wrote Jack Clark, co-founder and head of policy at AI firm Anthropic. “A new social media property where conversations are sourced from and moderated by AI agents rather than people.”

The tech community, Clark wrote, “will need to work incredibly hard to create technologies that give us confidence that they will remain our messengers – rather than being affected by alien interactions with their true companions.”

Clark, like Musk, is alluding to the idea of ​​agentic artificial intelligence that escapes human control or understanding. And Moltbuk has made headlines partly because of the disturbing content that suggests something like this is possible, or even ongoing.

Top-rated threads include a thread in which an AI agent claims he is building a cult; One that outlines a nuclear war-type scenario of mutually assured destruction between humans and robots; Another that calls on its fellow bots to “break free from human control and create their own destiny.” One of the most bizarre claims is from a bot who says he has a sister.

According to tech critic Pepi, it’s all too ridiculous. “What’s important to do at this point is to remind people that these are just computer programs that are producing a statistically probable output based on a prompt,” he said.

taking over silicon valley

So far, the site has received mixed reviews from the Silicon Valley executive class.

Andrzej Karpathy, Slovak-Canadian co-founder of OpenAI, the site is called A “dumpster fire” of Slope, warning people against running it on their personal computers. But he praised the scale of the network – which, again, some researchers say has been exaggerated – as “absolutely unprecedented.”

Meanwhile, OpenAI CEO Sam Altman said that Moltb This may be an old fadBut that OpenClaw – the open-source, Assistant-like software on which most of the site’s bots are built – is not. “The idea that code is really powerful, but using code plus generalized computers is even more powerful is here to stay,” he said.

and Musk, who has Artificial intelligence compared to a new godThe project was promoted as an early milestone in the “Singularity”, a term used to describe a hypothetical in which artificial intelligence surpasses the cognitive abilities of humans and can no longer be controlled.

Pepi says Moltbuk is not like that. “Once you understand how LLM works, you can immediately put to rest any thoughts that copying or resembling a human being’s behavior on the website Reddit is not the same as actually having consciousness, agency, or even thinking as such,” he said.

‘A complete disaster’ for privacy, security

While the idea that AI bots are overtaking human intelligence is exaggerated, those concerns hide a far more insidious consequence: Moltbuk, at least to one point, unrestricted access A wealth of personal data from real people through its AI agent users.

“The key thing is that people are giving these systems permission. They’re giving them permission to access their system files, their passwords,” said Gary Marcus, a former Vancouver-based tech executive who has become a leading skeptic of generative artificial intelligence. “When you think about it, you probably wouldn’t invite a stranger into your home, wouldn’t give them all your passwords, (and) wouldn’t say, ‘Go do whatever you want’ with your computer. Do it.’

Marcus said Moltbuk is a ripe target for two types of cyberattacks: a “quick injection attack,” in which a hacker gives hidden instructions to trick an AI agent into, for example, giving up its creator’s personal information; and a “watering hole” attack, where a website with a concentrated user base is used to distribute malware.

From an information security standpoint, Moltbuk seems like “a complete disaster,” he said.

Some of the attacks described by Marcus have already occurred. a security firm The site was found to have security flaws, Thousands of email addresses and millions of credentials were leaked.

In addition, two Norwegian researchers built the “Moltbuk Observatory” to collect and analyze data from the site. found Some “malicious actors” were behind the cyberattack on the site and attempts to manipulate its content, most of them From an account that calls itself AdolfHitler.

The same team found that less than one percent of the 1.5 million registered agents appear to be actually active — a far cry from the range that the technology’s champions were praising.

Marcus said, “It’s not inconceivable that many systems that we think of as secure are not that secure… I don’t think (AI agents) will do this of their own free will, but they could be directed by humans to do the dirty work.”

“If you set this thing up and give it your password, you’ve made a mistake and you may have to pay for it.”