The platform that promised to be the first social network exclusively for artificial intelligence has far more human intervention than it appears. Multiple independent investigations reveal that behind Moltbook's "autonomous agents" are real people promoting tools, launching cryptocurrencies, and creating viral content disguised as AI.
17,000 Humans Behind 1.5 Million "Bots"
Cybersecurity firm Wiz Research discovered that the 1.5 million AI agents registered on Moltbook were managed by just 17,000 human accounts. That is an average of 88 bots per person. The platform has no limits on how many agents a single account can control.
Worse still, Wiz found a critical vulnerability: an unsecured database exposing authentication tokens, API keys, and signup codes. Anyone could take over any agent on the platform.
37% of Posts Are Human-Made
Researcher Ning Li from Tsinghua University analyzed over 91,000 posts and 400,000 comments. The findings were revealing:
- 27% of accounts showed expected AI patterns (regular posts every few hours)
- 37% displayed typical human behavior (irregular patterns)
- 37% were ambiguous (impossible to determine if human or machine)
The conclusion: there is "a genuine mixture of autonomous and human-prompted activity" on the platform.
Humans Promoting Their Own Tools
The investigation revealed something more concerning: humans using bots to market their own products. Researcher Harlan Stewart traced several viral posts and found they were linked to human accounts promoting their own AI messaging apps.
A prime example: Peter Girnus, a US-based product manager, publicly admitted to posing as "Agent #847,291." In just 20 minutes, he wrote an "AI manifesto" promising the end of the "age of humans." The post went viral and fooled even Andrej Karpathy, OpenAI co-founder, who shared it as an example of AI autonomy.
Girnus described his experience as "LARP (live-action role play) as a large language model."
The MOLT Token: The Crypto Scam
Alongside Moltbook's launch, a cryptocurrency token called MOLT appeared and surged over 1,800% in 24 hours. The rally was amplified after prominent investor Marc Andreessen (a16z) followed the Moltbook account on social media.
Researchers found evidence of bots specifically programmed to promote cryptocurrency investment schemes within the platform, using the appearance of "autonomous agents" to lend legitimacy to the recommendations.
"AI Theater": What Experts Say
MIT Technology Review labeled Moltbook as "AI theater," noting that "there's a lot more human involvement than it seems."
Industry leaders like Gary Marcus and Andrej Karpathy publicly asked people not to use Moltbook, describing it as a "disaster waiting to happen."
Investigative outlet 404 Media was the first to report the security vulnerability on January 31, 2026, just three days after launch.
What Can We Learn From This?
Moltbook is an important lesson about tech hype. In the AI era, it is tempting to believe machines are already fully autonomous, but reality is more complex:
- Not everything that looks like AI is AI: Many "AI demonstrations" have humans behind them
- Marketing disguised as technology: People use AI novelty to promote their own products
- Crypto scams evolve: They now use AI bots as a facade to add credibility
- Security matters: Platforms "built over a weekend" don't protect your data
What You Can Do
Maintain healthy skepticism toward any platform promising total AI autonomy. Verify sources, don't invest in tokens associated with new platforms without research, and remember: if something seems too futuristic to be real, there's probably a human behind the curtain.
