Online comment increasingly sound synthetic, while bot traffic and astroturfing are overwhelming genuine conversations, opines one AI firm’s CEO.
Bots and algorithmic manipulation have made social media posts start to feel “fake”, according to the CEO of OpenAI, Sam Altman, who posted his views on 8 Sep 2025.
Altman has voiced concerns that many trending comments could actually be authored by non-human accounts — bots, paid astroturfers, or humans simply adopting the linguistic quirks of large language models. He argues that the “Extremely Online” crowd on platforms such as X and Reddit tends to drift towards group behavior patterns and lingo reminiscent of AI-generated text, further blurring the line between human and synthetic communication.
The skepticism is rooted in social media’s incentives, where, he noted, platforms and creators increasingly monetize engagement, and push users and algorithms alike to optimize posts for clicks and visibility.
This convergence incentivizes bot-like automation and astroturfing — sometimes paid invisibly by rival firms seeking plausible deniability — creating communities that spiral into hype cycles or toxic echo chambers. In the recent launch of his firm’s GPT 5.0 model, waves of critical posts had diminished positive sentiment, sometimes leaving observers unsure whether the backlash was organic or artificially amplified, Altman reflected.
Industry estimates suggest that over half of all internet traffic in 2024 was non-human, with large language models accounting for much of the surge. X’s own Grok bot estimated “hundreds of millions” of bots active on its service, though no figures were officially divulged.
The open lament is fueling speculation about OpenAI’s rumored plans for its own social platform. However, critics warn that such a network could still not escape bots, referencing research from the University of Amsterdam where bot-only networks quickly devolved into self-replicating cliques and echo chambers.
According to the post, the ‘authenticity crisis’ on social platforms is now pervasive, driven by advances in language models, aggressive engagement strategies, and widespread algorithmic activity that undermine trust in online conversations.
