OpenAI weighs ‘human-verification social network’…considers iris and Face ID to block bots

OpenAI is said to be considering building its own social networking service (SNS), putting the fight against bots front and center. Internally, the company is discussing ways to create a “platform where only real humans can participate,” using biometric authentication such as iris recognition or facial recognition.

According to Forbes on the 28th (local time), OpenAI is developing the SNS project privately with a small team, and is reported to have considered using Apple’s Face ID or an iris scan orb from World for user identity verification. The concept is understood to directly target structural problems at X (formerly Twitter), where bot accounts have proliferated.

Citing multiple sources, Forbes reported that OpenAI’s SNS is positioning “a network that only real humans can join” as its key differentiator. World is a blockchain project run by Tools for Humanity, which was co-founded by Sam Altman, CEO of OpenAI, and it scans irises to generate unique digital identity credentials.

While Meta’s Facebook and LinkedIn currently use phone number- and email-based verification, there are no known cases of adopting full-scale human verification using biometrics. Privacy advocates warn that if immutable biometric data such as iris information is leaked, it could pose serious risks.

How OpenAI’s SNS would integrate with its existing services remains unclear. However, sources said a likely model is one in which users use AI to create and share content such as images and videos themselves. No launch timeline has been set, and the development direction could change.

The bot-account issue is widely seen as a persistent challenge across the social media industry. X, in particular, is assessed to have seen bot proliferation worsen after Elon Musk’s acquisition, as large-scale headcount reductions weakened trust and safety teams. Musk said he removed about 1.7 million bot accounts during 2025, but the problem persists.

Altman has also publicly criticized X’s bot problem. Last year, he wrote on X, “AI Twitter and AI Reddit feel much more fake than they did just 1–2 years ago,” adding, “It feels like there are now a lot of accounts run by large language models.”