Sam Altman says that bots are making social media feel ‘fake’ 

X Enthusiastic and Reddit shareholder Sam Altman had a revelation on Monday: Bots made it impossible to determine whether messages on social media were really written by people, he posted.

The awareness came while reading (and sharing) some messages from the R/Claudode Subreddit, which praised OpenAi Codex. OpenAI launched the software programming service that takes the Claude code of Anthropic in May.

Recently, that subreddit has been so filled with messages from self-proclaimed code users who announce that they have moved to Codex that a Reddit user even joked: “Is it possible to switch to Codex without placing a subject on Reddit?”

Because of this Altman wondered how many of those messages were from real people. “I had the strangest experience to read this: I assume it is all fake/bots, although in this case I know that Codex growth is really strong and the trend here is,” he confessed to X.

He then analyzed his reasoning. “I think there are a bunch of things going on: real people have picked up quirks of llm-speak, the extremely online crowd drifts Together in very correlated ways, the hype cycle has a very ‘it’s optation so on/we’re so-so-so-so-so-so-so-so-so-so-so-so-so-so-so-so-so-so-so-so-so-wee press Juicing Engagement and the Related Way that creatoration works, other Companies have astroturfed us so I’m extra sensitive to it, and a bunch more (including probable some bots) “.

To decode that a bit, he accuses people to start sounding like LLMS, although LLMS – led by OpenAi – was literally invented to simulate human communication, up to the EM dashboard. And the models of OpenAi are certainly trained on Reddit, where Altman was a board member until 2022, and was announced as a large shareholder During the IPO of the company last year.

He makes a valid point that Fandoms, led by extreme, always on social media users, tend to behave in strange ways. Many groups can be flooded anywhere in Hatefests if they are flooded by those who express frustrations for their brothers.

Altman also throws an excavation to the stimuli when social media sites and makers rely on involvement in making money. Reasonable.

But then Altman confesses that one of the reasons why he thinks the pro-opai messages in these subreddit can be bots, because OpenAi is also ‘astroturfed’. This usually includes messages from people or bots who are paid by the competitor, or paid by a third -degree contractor, which gives the competitor plausible denial.

We have no proof of astroturfing (although it is possible). But we have seen how OpenAi Subreddits the company turned on After the GPT 5.0 had released. Instead of praising the believers about the new model, many angry posts were voted. People went to Reddit and X to complain about everything, from GPT’s personality to how it was burned by credits without tasks.

A day after the bumpy release, Altman did a Reddit Ask me. Session about R/GPT in which he rolled out known problems and promised changes. The GPT breddit has never fully recovered its previous level of love, with Users still post regularly About how much they don’t like the changes with GPT 5.0. Are they human? Or are they, as Altman seems to imply, in one way or another fake?

Altman suspects: “The net effect is somehow AI Twitter/AI Reddit feels very fake in a way that really doesn’t have it a year or two ago.”

If that is true, who is it then? GPT has led models to become so good in writing that LLMS has become a plague, not only for social media sites (who have always had a BOT problem) but to schools” journalismAnd The courts.

Although we do not know how many Reddit messages have been written by bots, or are fictional accounts by people who use LLMS, it is probably a considerable number. Data Security Company Imperva reported that More than half of all internet traffic in 2024 Was non-human, largely due to LLMS. X’s own Bot Grok says: “The exact figures are not public, but 2024 estimates suggest hundreds of millions of bots on X.”

Several cynics have suggested that Altman’s lamentation was his first trip to the rumors of Marketing OpenAI’s rumors social media platform. In April, the Verge reported that such a project to accept X and Facebook was in the earliest phases. This product may or may not exist. Altman may or may not have had left motives to suggest that social media are too fake nowadays.

But aside motifs, if OpenAI is planting a social network, what are the chance that it would be a bone -free zone? And, funny enough, if the opposite and people would banish people, the results would probably be no different. LLMS hallucinating not only facts, but when researchers from the University of Amsterdam built a social network that consisted entirely of Bots, they discovered that the bots soon Formed cliques and ultrasound rooms for themselvesat.