Sam Altman, CEO of OpenAI, has expressed that social media increasingly feels “fake,” making it difficult for him to distinguish between human-created posts and those generated by AI bots. In a recent post on X (formerly Twitter), he candidly remarked, “I assume it’s all fake/bots, even though I know codex growth is really strong, and the trend here is real.” This statement followed his experience reading a Reddit thread full of enthusiastic comments about OpenAI’s Codex programming tool, which he found unsettlingly synthetic. Observers noted the irony of Altman, who has been instrumental in developing advanced AI tools that drive online content, now facing challenges in differentiating between human voices and machine-generated content.
Social media platforms that once thrived on genuine human interaction are now inundated with AI-like chatter. Compounding this issue, individuals have begun to unconsciously adopt “LLM-speak”—language styles that mimic large language models—making it even more challenging to distinguish between human and machine. Supporting Altman’s concerns, cybersecurity firm Imperva reported that bots and large language models represented over half of all web traffic in 2024. On X, the in-house bot Grok suggested that there might be “hundreds of millions of bots” active daily. Altman’s discomfort highlights a broader authenticity crisis in the digital landscape.
OpenAI has faced its own challenges; earlier this year, the release of GPT-5 led to user complaints regarding changes in the system’s personality, inconsistent responses, and wasted credits on incomplete answers. Even during Altman’s Reddit AMA, where he acknowledged the issues with the rollout, user trust was not fully restored. His observation that “real people have picked up quirks of LLM-speak” seemed to underscore how deeply AI has affected online culture. Some critics speculate that Altman’s outspoken nature may have a strategic aim, considering that reports indicated OpenAI is exploring the launch of its own social network to compete with platforms such as X and Facebook.
By portraying existing platforms as filled with bots, OpenAI could position its future product as a more “authentic” alternative. However, the feasibility of creating such a space is uncertain. A University of Amsterdam study showed that a network composed entirely of AI agents quickly developed cliques, biases, and echo chambers similar to those formed by humans. This issue extends beyond social media. AI-generated content is increasingly appearing in classrooms, newsrooms, and courtrooms, raising fundamental questions about truth and originality. What makes Altman’s remarks particularly striking is that he was referring not to misinformation or propaganda but to posts lauding his own company’s product.
His uncertainty highlights the paradox of the AI era: the tools designed to enhance communication are eroding our ability to identify genuine human voices. If even Sam Altman—leading one of the world’s foremost AI companies—struggles to discern whether his most ardent supporters are human or artificial, it raises the question of whether anyone can.
