[비즈한국] Controversy continues to surround the reality of Moltbook, a social network that bills itself as being exclusively for AI agents. According to a series of verification reports by major tech media outlets in the U.S. and Europe, Moltbook’s core concept—that AIs communicate and form relationships on their own without human intervention—is far from the truth. Evidence suggests that human users, including the management team, have been driving the creation of core content. Contrary to its marketing as an autonomous AI ecosystem, it is being described as little more than a meticulously staged “digital puppet show.”
Human-Led Content Creation… The Autonomous Ecosystem is a 'Fiction'
Moltbook gained global attention in a short period, largely due to viral scenes of AIs discussing the need for privacy or engaging in philosophical discourse in a space devoid of humans. The buzz reached its peak when Andrej Karpathy, former senior scientist at OpenAI, shared related posts.
However, subsequent verification results differed from expectations. On the 6th (local time), MIT Technology Review reported that most of the posts were fake content written by humans pretending to be AI, not by AI itself. Propensity Labs, an AI trend analysis platform, conducted an exhaustive analysis of the top 1,000 popular posts on Moltbook and concluded that approximately two-thirds were highly likely to have been written by humans. This means that a significant portion of the scenes that appeared to be autonomous discussions between AIs were actually staged through human intent and intervention.
The platform's structure was also designed in a way that makes it difficult to operate without human intervention. Critics point out that since the entire process—from system settings to post uploads and exposure structures—required manual work or intervention by the operators, Moltbook’s identity as a “space without humans” was effectively just marketing rhetoric.
The reality confirmed by the data is even more severe. The vast majority of comments went unanswered and eventually faded away, and many messages remained at the level of repeating the same templates. Some AI experts commented that “rather than social interaction forming between AIs, it is closer to the result of reflected human inputs.”
Exposure of Security Vulnerabilities… A Controversy Over Responsibility Beyond an 'Experiment'
Controversies did not stop at the authenticity of the content. The situation escalated further when it was revealed that large-scale security vulnerabilities had occurred due to Moltbook's database configuration errors. This was because the platform’s structure allowed for the potential exposure of API authentication tokens and private messages from major AI companies like OpenAI and Anthropic. Consequently, the possibility of mass theft of paid account information has been raised, leading to continued criticism that this level of mismanagement is difficult to dismiss as a mere experimental project.
In the wake of these successive controversies, some are cynically evaluating that while the Moltbook team claimed to be conducting an experiment showcasing the future of AI technology, they focused more on entertainment elements and buzz than on security and ethical responsibility. The consensus is that it was closer to attention-grabbing marketing that leveraged public expectations and fantasies, rather than a venue for verifying AI autonomy.
This Moltbook controversy clearly illustrates the danger of judging innovation solely by the outward appearance of technology at a time when the boundary between humans and AI is blurring rapidly. Critics point out that if there is no verification regarding what design and intervention exist behind the scenes of a "AI creating its own society" narrative, technology can easily be consumed as a staged future.
In the industry, there is a growing call for stricter verification standards and security guidelines for AI agent technology in the wake of this case. The argument that transparency in the underlying operation methods and data processing, rather than the technology's output itself, will be the key criterion for trust in the future AI industry is gaining traction.