At the end of January 2026, a report from Forbes sent ripples through the tech circle: OpenAI is assembling a small internal team to develop a brand-new social networking product. Different from traditional social platforms, the core concept of this product is only four words — Prove You Are Human.
According to the report, OpenAI is evaluating multiple biometric verification solutions, including Apple’s Face ID and the Orb iris scanning device from World Network (formerly Worldcoin). The news sent Worldcoin’s token WLD surging more than 25% within hours, with trading volume jumping over tenfold. The violent market reaction reflects a profound signal: today, as AI is deeply embedded in human society, iris recognition is moving from a "niche technology" to a strategic high ground of "infrastructure-level".
I. An Overlooked Paradox: Why Do Creators of AI Need to "Exclude AI"?
The most intriguing part of this story lies in its inherent contradiction. OpenAI is the most powerful AI content generation engine today; ChatGPT can write indistinguishable articles, comments and dialogues in seconds. Yet it is precisely this capability that has made the proliferation of bots on social networks an unavoidable governance challenge.
The X platform (formerly Twitter) has long been plagued by automated accounts. In January 2026, X was forced to update its API rules and ban a number of "InfoFi" projects that drive posts through economic incentives, as these projects spawned massive amounts of AI-generated spam content. However, blocking is only a temporary fix. As AI-generated text, images and even videos become increasingly hard for the human eye to distinguish, platforms must answer a fundamental question: Is the person on the other side of the screen really human?
OpenAI’s choice is philosophical in nature — using biometrics to draw the boundary between humans and machines. And among all biometric features, it has set its sights on the iris.
II. Why Iris? A Technical Contest Over "Entropy"
Tools for Humanity (the developer of World Network) once gave an incisive explanation in a technical blog: the iris was chosen because it carries more randomness and complexity than fingerprints or faces, making it far harder to forge.
The technical meaning of this statement is worth elaborating. In biometric recognition, entropy is a key indicator measuring how much independent identity information a feature can carry. Studies show that iris coding can generate more than 200 bits of independent entropy, meaning it can theoretically distinguish every individual among the 8 billion people on Earth. In comparison, the entropy of fingerprints is a fraction of that of the iris, and the accuracy of face recognition is roughly four orders of magnitude (i.e., 10,000 times) lower than iris recognition.
More importantly, the iris is an internal human organ, naturally protected by the cornea. It does not wear or change due to labor or trauma like fingerprints, nor does it alter drastically with age, makeup or lighting conditions like facial features. The iris texture of a healthy adult remains almost unchanged throughout life. Even identical twins have completely different iris textures — because iris formation stems from random processes during embryonic development, independent of DNA.
These traits make the iris the only biometric feature capable of reliable unique verification at a global scale (1:N matching). As noted in the World Network white paper, the False Match Rate (FAR) of iris recognition can be as low as 2.5×10⁻¹⁴ (1 in 40 trillion), a precision unmatched by any other biometric feature.
III. From Social Networks to Digital Identity: The Strategic Upgrade of Iris Recognition
OpenAI’s exploration goes far beyond solving the bot problem on social platforms. It reveals a grander industrial trend: on the eve of AI Agents’ large-scale entry into human society, "human-machine distinction" will become the most fundamental trust mechanism in the digital world.
Imagine a scenario: three years from now, your medical appointments are made by AI agents, your tax filings are processed by AI agents, and over half of the content on your social media is AI-generated. In such a world, how do you prove you are not a machine? Passwords can be cracked, faces can be deepfaked, fingerprints can be copied — yet the iris, with its irreplicable high-entropy traits and lifelong stability, may become the ultimate answer to this question.
This perception is spreading from Silicon Valley to the world. Apple has introduced the iris-based Optic ID authentication system in Vision Pro; India completed an iris database covering 1.3 billion people years ago (the Aadhaar project); the UAE, the United States, the UK and other countries have deployed iris recognition clearance systems in more than 50 international airports; the Ministry of Public Security of China has also piloted the construction of a national iris database since 2019.
The exploration by Worldcoin/World Network is even more radical — it seeks to establish Proof of Personhood based on iris for every person on Earth, having completed Orb iris verification for over 12 million people so far. Although the project has sparked much controversy over data privacy, its underlying logic has gained increasing recognition: in an era of exponential AI growth, biometrics — especially the iris — may be humanity’s most reliable credential to prove its existence.
IV. China’s Answer: 14 Years of Accumulation from Algorithms to Chips
China is not a latecomer on the track of iris recognition.
Founded in 2011, WuHan Homsh Technology Co., Ltd. has long focused on the independent R&D of core iris recognition algorithms and dedicated chips. It is one of the very few companies worldwide that achieves full-stack independent control across "algorithm – chip – module – terminal – solution". The company owns the self-developed PhaseIris series of iris recognition algorithms, which adopt an iris code length of up to 64Kb. The single-eye recognition accuracy reaches 1 in 10 million, and binocular recognition accuracy reaches 1 in 1 billion. Under the same hardware conditions, the algorithm speed is nearly 5 times faster than the classic Daugman algorithm.
More strategically, Homsh successfully developed and taped out the world’s first dedicated ASIC chip for iris recognition — the Qianxin QX8001 — in 2018. Measuring only 3.2×2.8mm², this chip compresses iris encoding speed to under 0.05 seconds, achieves a matching speed of 90,000 times per second, and consumes only 0.1W on average — far lower than the 6–7W power consumption of international competitors. The chip adopts a fully hardware-based AES-256 encryption algorithm, ensuring the security of iris template data at the underlying level.
This technical path of "solidifying algorithms into hardware" precisely aligns with the core demand for the large-scale application of iris recognition. Over the next decade, the global number of IoT devices is expected to reach hundreds of billions. Every smart terminal requiring identity authentication imposes strict requirements on cost, size and power consumption. Pure software solutions rely on general-purpose processors (CPU/GPU) and struggle to meet the needs of large-scale deployment in these three dimensions. Dedicated chips, however, can reduce the per-unit cost to below $1 after mass production, providing economic feasibility for the popularization of iris recognition.
To date, Homsh has accumulated more than 280 intellectual property rights, participated in the formulation of multiple national and industry standards for iris recognition, and built a complete product system covering iris collectors, iris access controls, smart iris locks, and iris cloud computing platforms. Its products are applied in public security, smart cities, energy management, financial security, entry-exit control and other fields. The company is advancing the mass production of the new-generation Qianxin Q80 chip, targeting emerging markets such as metaverse/VR devices, security electronics and smart vehicles, aiming to become a key supplier in the field of human-computer interaction semiconductor chips in the AI era.
V. In Closing: The Body Is the Password, the Iris Is the Trust
The move of OpenAI + Worldcoin is not an isolated event, but an inevitable extension of the logic of technological evolution. As AI has grown powerful enough to simulate all human external expressions — text, voice, face and even behavioral patterns — humanity has to trace back to a deeper biological essence to find that "unforgeable" anchor.
The iris, a circular tissue merely 12 millimeters in diameter, carries everyone’s unique "biological password". Shaped by chaotic random processes during embryogenesis, it is not fully governed by genes, unchanged by time, and undamaged by external forces. Among all identification methods invented by humans, it comes closest to an "unreplicable proof of existence".
We stand at a historic crossroads. AI is redefining "what is real", while iris recognition answers an older question — "Who are you?". In this technological competition concerning the underlying trust of digital civilization, enterprises mastering core algorithms and chip capabilities will hold the power to define the future.
About Homsh
WuHan Homsh Technology Co., Ltd., founded in 2011, is a national high-tech enterprise focused on the R&D of core iris recognition algorithms and dedicated chips.
The company owns the PhaseIris iris recognition algorithm and Qianxin series ASIC chips with fully independent intellectual property rights, and has built a complete industrial chain covering "chip + module + terminal + solution".
Homsh has filed for more than 280 intellectual property rights in total and is a participant in the formulation of national standards for iris recognition. It is committed to becoming a global leading provider of iris recognition technology and chips in the AI era.
