Over the past few years, millions of people, mostly in the Global South, have accepted an enticing offer: the promise of free cryptocurrency—25 ethereum-based Worldcoins, to be airdropped into a Worldcoin Wallet app—in exchange for having their irises scanned by a fancy silver orb. The Worldcoin project is a brainchild of Sam Altman, formulated in 2019 just as he was stepping sideways from his position as President of Y Combinator, the Silicon Valley start-up incubator, to become ceo of Openai, the for-profit not-for-profit company supposedly dedicated to making sure that artificial intelligence would be ‘safe and beneficial’ for humankind. With billions from Microsoft, the computer scientists at Openai trained their deep-learning Large-Language Models (llms) to mimic the human use of language by crunching loot from Wikipedia and wider scrapes of internet content, producing dall-e—an image prompt generator—as well as the Generative Pre-trained Transformer systems (gpts), most famously Chatgpt, released in November 2022. This family of models could not only carry on a conversation but confidently filibuster on just about anything, disregarding facts and sources.

‘The global economy belongs to everyone’ is Worldcoin’s slogan. Altman, the project’s co-founder alongside ceo Alex Blania, has described its goals in lofty terms: not only a decentralized identity and financial network, owned by everyone, but a first step towards providing a global Universal Basic Income, sharing the wealth that ai would generate with those whom it would make redundant; or as the Worldcoin website puts it, a ‘more human’ internet, a ‘more human’ blockchain protocol and a ‘more human’ economic system, accessible to all.footnote1 In fact, as Altman and Blania made clear, Worldcoin is supposed to solve a problem that Openai and its rivals at Google, Amazon and the rest have created: how to distinguish online between digital entities and real human beings, now that ai systems can sail through captcha tests, recognize images and generate plausible text? Biometric data, Blania affirmed, would be the ultimate ‘proof of personhood’; an upgraded Turing test in the age of ai. Worldcoin was getting out ahead of the competition, with its proposition of trading iris scans for ‘fiat’ crypto-currency tokens. The company described the operation as ‘the biggest onboarding into crypto and Web3’ to date.footnote2

In their investigation of Worldcoin’s procedures, researchers at the mit Technology Review uncovered deceptive marketing practices, exploited workers, failure to secure informed consent and the collection of more biometric information than was acknowledged—as well as glaring inequalities between data collectors and data providers. The researchers were puzzled at first about the actual goal of the data collection, which Worldcoin representatives would only describe in the vaguest terms, while keeping their code and technology secret. Eventually they came to the conclusion that crypto and ubi were just an afterthought to incentivize people in countries outside the realm of data-privacy regulation to relinquish their biometric data, taking advantage of the difficult financial situation of the pandemic. The chief goal was to train Openai’s neural networks on this data. The scans turned out to be a scam. What appeared to be a crypto-currency onboarding tool was in fact a large-scale data-collection operation.footnote3

Twenty years ago, the Harvard anthropologist Michael Herzfeld coined the term crypto-colonialism to describe territories that are not colonies in the standard definition of the word, but suffer more indirect forms of oppression. Herzfeld wrote of ‘the curious alchemy’ whereby certain countries—he mentioned Greece and Thailand—‘were compelled to acquire their political independence at the expense of massive economic dependence’, this relationship being articulated in the ‘iconic guise of aggressively national culture fashioned to suit foreign models.’ Such countries were and are living paradoxes: ‘they are nominally independent, but that independence comes at the price of a sometimes humiliating form of effective dependence.’footnote4

Herzfeld wrote these lines long before Satoshi Nakamoto invented Bitcoin. But his approach may help to illuminate some of the deeper roots of the Worldcoin phenomenon, for ‘crypto-colonialism’ neatly maps the geopolitical relations that underlie the data thefts and cryptoscams carried out by ai corporations. The Worldcoin example connects two technologies, ai and blockchain, whose origins are quite distinct, though both have been subjects of techno-hype. A cryptocraze took off in 2021, during the pandemic lockdowns, fuelled in part by young American white-collar workers gambling their furlough cheques in online investment. In 2022, just as this bubble crashed, the exponential spread of generative machine-learning applications created another wave of tech speculation. Another year on, and multisystemic crises—wars, climate change, inflation, energy impasses, inequality—are interwoven with hype around llm-powered chatbots, image-generation tools and, most recently, boardroom drama at Openai, where Altman and the staff saw off an attempted coup by another board faction.footnote5

In the Worldcoin project, machine-learning and blockchain technologies come together in peculiar new ways. In addition to a potential scam to extract biometric data from vulnerable people in exchange for pseudo-money, the performative staging of Worldcoin involves a tangle of social and technological relations that range from magic ritual to machine learning, from the exchange of gazes to DeFi wallet apps. The silver orb seems to represent some higher knowledge and to connect the people who stare into it with some mythical planetary community. This aspect is documented in the promotional photos of Worldcoin scanning sessions, where the orb and the person being scanned appear locked in some kind of mutual gaze, oblivious of their surroundings, as if in a twisted endurance performance by an automated Marina Abramović.

As the scannee stares into the orb, its ‘contactless doppler radar detection’ supposedly records heartbeats, breathing and other vital signs, as well as iris and facial data.footnote6 Different modes of sensing confront and interpenetrate each other here, as the horizontal eye-level encounter characteristic of human communication is extended into the machinic realm. Seeing and scanning, observation and perception, concentration and recording: they create a sphere of communication and exchange that might be called ‘common sensing’, linking crypto fintech and a machine-learning quasi-monopoly to populations hard hit by pandemic conditions, in a relationship marked by deception and one-sided dependency. It is widely recognized that in a digital context, ‘sharing’ is usually a polite-slash-creepy expression for the expropriation of one party, the less well-off, by another. Here, the space shared between human and technical gadget turns out to facilitate cheap data acquisition.