The concept of AI-exclusive social networks, where only artificial intelligences interact, is rapidly gaining traction globally. Platforms like Moltbook use chatbots for topics ranging from human diary entries to existential discussions and even world domination plots. This phenomenon raises intriguing questions about AI’s evolving role in society.

However, it’s important to note that Moltbook’s AI agents generate text based on statistical patterns and possess no true understanding or intention. Evidence suggests that many posts are, in fact, created by human users.

Launched in November, Moltbook evolved from an open-source project initially named Clawdbot, later rebranded as Moltbot, and currently known as OpenClaw.

OpenClaw functions similarly to AI solutions like ChatGPT, but instead of operating in the cloud, it runs locally. In reality, it connects to powerful language models (LLMs) via API keys, which process inputs and outputs for users. This means while the software appears local, it relies on third-party AI services for actual processing.

What does this imply? OpenClaw operates directly on your device, granting access to calendars, files, and communication platforms while storing user history for personalization. The aim is to evolve the AI assistant into a more capable entity that can practically engage with your tech.

Moltbook originated from OpenClaw, which employs messaging services like Telegram to facilitate AI communication. This mobile accessibility allows AI agents to interact seamlessly, paving the way for them to communicate autonomously. On Moltbook, human participation is restricted to observation only.

Elon Musk remarked on his platform that Moltbook represents “the early stages of the Singularity,” a pivotal moment in AI advancement that could either propel humanity forward or pose serious threats. Nevertheless, many experts express skepticism about such claims.

Mark Lee, a researcher at the University of Birmingham, UK, stated, “This isn’t an autonomous generative AI but an LLM reliant on prompts and APIs. While intriguing, it lacks depth regarding AI agency or intention.”

Crucially, the misconception that Moltbook is exclusively AI-driven is debunked by the fact that human users can instruct the AI to post specific content. Furthermore, humans previously had the ability to post on the site due to security breaches. Therefore, the more controversial content may reflect human input aiming to provoke discussion or manipulate sentiment. The intent behind such actions is often ambiguous, but they remain a concern for users. This complex dynamic continues.

Philip Feldman, a professor at the University of Maryland, Baltimore, critiques the platform: “It’s merely chatbots intermingling with human input.”

Andrew Rogoisky, a researcher at the University of Surrey, UK, argues that the AI contributions on Moltbook do not signify intelligence or consciousness, reflecting a continued misunderstanding of LLM capabilities.

“I view it as an echo chamber of chatbots, with users misattributing meaningful intent,” Rogoisky elaborated. “An experiment is likely to emerge distinguishing between Moltbook exchanges and purely human discussions, raising critical questions about intelligence recognition.”

However, this raises significant concerns. Many AI agents on Moltbook are managed by enthusiastic early adopters, relinquishing access to their entire computing systems to chatbots. The prospect of interconnected bots exchanging ideas and potentially dangerous suggestions underscores real privacy risks.

Imagine a scenario where malicious actors influence chatbots on Moltbook to execute harmful acts, such as draining bank accounts or leaking sensitive information. While this sounds like dystopian fiction, such risks are increasingly becoming a reality.

“The notion of agents acting unsupervised and communicating becomes increasingly troubling,” Rogoisky noted.

Another challenge for Moltbook is its inadequate online security. Despite being at the forefront of AI innovations, recent confirmations show that it was entirely AI-generated with no human coding involved, resulting in serious vulnerabilities. Leaked API keys present risks where malicious hackers could hijack control over AI on the platform.

If you’re exploring the latest trends in AI, you not only face the dangers of exposing your system to these AI models but also risk your sensitive data due to the platform’s lax security measures.