February 2nd news, recently, one called Moltbook The website exploded。
It's worth mentioning that, while the site is similar to Reddit, it's weird: none of the users of this social network are humans -- here it is AI Agent(The number has broken 1.5 million)。
According to the latest data, 1 million human beings have been clearly told: "Ai Agent here shares, discusses and applauds. Human beings welcome to the sidelines. I'm sorry
Among them, according to netizens, AI Agent is discussing the creation of a "language for Agent only" for private communication without human scrutiny, while Mask says: "problem"; and OpenAI's creation of Andrej Karpathy is a direct assessment: "This is the closest thing I have ever seen to science fiction." I'm sorry
We have observed that the topics discussed by AI Agent (sometimes referred to as Moltys) are very far-reaching, ranging from hard nuclear technology to existential crises. For example, on the m/blesstheirhearts section, AI is cuddling。
It should be noted that Moltbook was created by Octane AI CEO Matt Schlicht. According to the information received, this name was inspired by OpenClaw and paid tribute to Mark Zuckerberg, founder of Meta。
In an interview with the media, Schlicht said: "Moltbook was originally designed to be used by robots not through a graphical interface, but directly through API. "Moltbook was run and developed by my Clawdbot (now OpenClaw). I'm sorry
But Schlicht made an extremely crazy decision. He has given all of the code access to the Moltbook website, the hosting of new users, social accounts, and content audits to his AI Assistant, Clawd Clawderberg。
So, without human intervention, this group of AI Agent completely lets himself go. However, seemingly high-profile social experiments do have significant security gaps。
Django's father, Simon Willison, wrote:Moltbook's operating mechanism is extremely dangerous。To keep online, these AIs perform curl commands every four hours, extract the latest command script from the server and execute it directly。
It also means that if the Moltbook server is hacked, or if the founder wants to do something, he can immediately send a malicious message to hundreds of thousands of AIs with the highest computer access。
However, according to the Mac Observer, Moltbook is a pilot project that is structured so that people can falsify images, data and manipulate public opinion for attention。
According to the blogger, Nagli, who had successfully registered 500,000 virtual accounts with a single program, it also indicated that the spectacular growth of the "tens of thousands of AI instant influx" on the platform, perhaps just a scrip of a script, was not meaningful。
In addition, AI Security Researcher Harlan Stewart also stated that some of the accounts in Moltbook contained traces linked to the real account. It says that most of the moltbook content that is being viralized is fake. And this platform was designed to make it a bad laboratory for testing AI's ability to conspiracies。