AI is making it increasingly cheap to create "fake humans," yet proving your humanity grows ever more difficult: The identity business from Dead Souls to iris and palm prints
Author: Thejaswini M A, The Token Dispatch
Compiled by: TechFlow
TechFlow: In his 1842 satirical novel Dead Souls, Gogol tells the story of purchasing lists of deceased serfs to use as paper assets. On today’s internet, similar issues are resurfacing as artificial intelligence can cheaply and batch-produce fake virtual identities. Determining the exact monetary value of proving a real human exists behind a screen is shifting from a theoretical question to a mechanism with market pricing. This article outlines the technical pathways and economic logic behind CAPTCHAs, black-market SMS verification services, proof-of-work, Worldcoin’s iris scanning, and palm vein recognition.
The Dead Souls Problem in the Digital World
A well-dressed stranger arrives in a provincial Russian town and begins purchasing lists of serfs who have died but have not yet been removed from the official census records.
This is the plot of Gogol’s 1842 novel Dead Souls. The scheme made sense at the time because Russian landowners were taxed per serf under their name, and census updates occurred only every few years. Landowners wanted to shed the tax burden of deceased individuals, while buyers could hold these nominal heads of household and apply for loans from the state bank.
The logic history provides is clear: Any system that distributes benefits per head without being able to verify the authenticity of the entity will eventually become flooded with fake accounts.
In the first few decades of the internet, being someone online was completely free; pretending to be multiple people was also free.
Websites allow registering multiple email addresses, social platforms permit pseudonyms, and the blockchain world allows one person to generate tens of thousands of wallet addresses in an afternoon. Once a system grants each account one voting vote, or airdrops an equal share to each wallet, those capable of creating accounts at the lowest cost will claim the largest portion.
Thus, human identity verification is not just an authentication technology, but an economic one. Since the internet requires proof of human identity, the key question becomes: What is the price of this proof?
Economic Costs of Black-Market Code Provision and CAPTCHAs
Researchers at Cambridge University have long tracked the costs of purchasing SMS verification codes, which remain one of the most common anti-counterfeiting defense lines today.
These prices fluctuate constantly with market demand. For example, on December 19, 2025, obtaining a verification code for Telegram cost $0.34, Facebook $0.21, and WhatsApp $0.25. The bulk manufacturing of fake accounts has itself formed an independent supply and demand market.
Figure: Black-market prices for SMS verification codes and fake account generation
A common method for generating fake accounts is using SIM cards. In September 2025, the U.S. Secret Service seized 300 SIM server clusters and 100,000 prepaid SIM cards. The involved syndicate used this hardware to batch-register tens of thousands of social and payment accounts.
Figure: Source cotsi
Another cost lies in the time invested by society as a whole. According to relevant research estimates, global internet users spend approximately 500 person-years annually solving CAPTCHA challenges such as identifying distorted letters or selecting traffic lights. If calculated using the U.S. minimum wage, the annual time value lost to verifying CAPTCHAs amounts to roughly $1.3 billion.
Even so, automated scripts and bots are learning to bypass these tests. In 2024, some research teams achieved a 100% machine-breaking rate against mainstream graphical CAPTCHAs using multimodal vision models.
Crypto Networks’ Approach: Stopping Head Counts
The crypto industry calls this type of fake account attack a Sybil attack, named after a classic 2002 paper by Microsoft researcher John Douceur.
Satoshi Nakamoto’s solution when designing Bitcoin was to simply stop counting heads.
Nakamoto knew that voting based on IP addresses wouldn’t work, since one person could easily rent a large number of proxy IPs. Bitcoin ties voting power to computational hash rate, where miners consume electricity to calculate hashes as proof of participation—a mechanism known as Proof of Work (PoW). Subsequent Proof of Stake (PoS) further bound governance rights to staked capital.
Whether computing power or capital, both are tangible scarce resources that cannot be infinitely replicated through empty account registrations. However, in scenarios requiring benefit distribution per genuine individual, purely capital-weighted systems still cannot solve the requirement of one-person-one-vote.
Worldcoin’s Orb Iris Solution and Dedicated Hardware Bottlenecks
World (formerly Worldcoin) attempted to return to the route of directly identifying humans, adopting a custom hardware device called Orb.
Orb captures images of the user’s face and iris via infrared cameras, confirms the individual is a genuine human with no duplicate registration, and then generates a unique digital identity credential. Official data shows that over 18 million people have completed verification so far.
World’s technical logic is that dedicated biometric hardware can provide extremely high human uniqueness and anti-counterfeiting capabilities. If standard smartphone cameras could capture irises, hackers could inject spoofed digital video streams using fake cameras; to prevent such injection, World must manufacture Orb hardware itself, strictly controlling sensors, lenses, and lighting environments.
Figure: Source world
But dedicated hardware also brings structural deployment challenges: the higher the precision of the device, the higher the thresholds for production, transportation, and coverage. Regarding data storage, World adopted a Secure Multi-Party Computation (SMPC) approach, splitting users' iris codes into encrypted fragments and distributing them across servers belonging to different institutions to prevent any single centralized entity from holding the complete biometric database.
However, regarding security dimensions, decentralized biometrics still faces calculations of attack costs. Attacking Bitcoin requires purchasing massive amounts of electricity, attacking PoS chains requires buying expensive tokens, but in biometric systems, if the value of airdropped tokens in a certain impoverished region exceeds a local resident's daily living cost, hackers can often organize intermediaries to use small amounts of cash to induce real residents to undergo offline eye scans, then resell the credentials for arbitrage.
Technical Evolution of Palm Print and Vein Recognition
Beyond iris scanning, palm-based biometric solutions are also accelerating commercial adoption. Amazon launched Amazon One palm-print payments for physical retail checkout, and Tencent WeChat introduced WeChat Palm Pay, deploying it at rail transit gates and convenience stores.
Palm recognition primarily combines two features: the epidermal palm print texture, and the subcutaneous venous vascular distribution structure illuminated by near-infrared light. Since subcutaneous veins exhibit specific infrared absorption characteristics only during live blood flow, faking them is significantly harder than forging standard flat-face photos.
In terms of user experience, extending a palm offers better psychological comfort and public health acceptance than pressing eyes against a metal sphere. Users do not need to remove sunglasses or bend down to align with a camera, making the process more aligned with daily habits when passing through gates or checkout counters.
Figure: Social media discussions on the security of palm-print authentication
Yet its core trade-off mirrors World’s: to achieve an extremely high level of anti-counterfeiting security, dedicated infrared vein sensors are required, making widespread adoption impossible relying solely on inexpensive standard smartphone front-facing cameras.
Conclusion
From the paper serfs of Dead Souls to today’s SMS black markets and automated bots, as long as systems contain mechanisms that distribute benefits per head, the industry of manufacturing fake identities will always persist.
In an era flooded with synthetic content and AI agents, the boundary for identifying genuine humans is gradually retreating from software code to biological entities. Whether iris or palm print, how to balance the security barriers of dedicated hardware with the convenience of global rollout remains the core challenge that the next phase of decentralized identity systems must answer.