If the U.S. government fears China’s progress in artificial intelligence, why is it giving Beijing a treasure trove of highly valuable crypto data with which to train its machine-learning models?
That’s the rhetorical question that developer Anish Mohammed left with me when we caught up at the NEARCon gathering in Lisbon this week.
It gave me a new, rather alarming perspective on the anti-privacy, self-censoring crackdown that’s playing out in crypto circles following the U.S. government’s sanctioning of Ethereum-based transaction mixer Tornado Cash.
I and many others were making the positive case long before the Tornado Cash action that the U.S. and its allies would gain a geopolitical advantage if they allowed the expansion of decentralized, privacy-protecting, open-access crypto protocols.
The idea is that Western society can unleash a “killer app”: the enshrining of individual rights into code, enticing the world’s digitally-mobile citizens to vote with their pocketbooks in favor of Western monetary systems that encourage open-access and privacy over authoritarian control and surveillance. The model would win, we said, because dictatorships and single-party states cannot match it without losing control over their money and thereby sowing the seeds of their own downfall.
You’re reading Money Reimagined, a weekly look at the technological, economic and social events and trends that are redefining our relationship with money and transforming the global financial system. Subscribe to get the full newsletter here.
Until my conversation with Mohammed, the chief scientist at privacy-protecting Panther Protocol, the failure to embrace that open approach to crypto policy was a major source of my frustration with the United States’ draconian regulation of crypto projects, where users are compelled to identify themselves, comply with “accredited investor” requirements and meet excessive reporting demands from government authorities. All of that activity, I’ve argued, defeats crypto’s decentralization principles and diminishes its potentially vast value to society.
But, sadly, I now have a more ominous take on the problem. It seems we’re not only relinquishing the opportunity to use the “carrot” of privacy to out-compete China, we’re also handing it a “stick” with which to subjugate us and accelerate its march toward AI dominance.
Data is the fuel of the AI economy
We have reached a stage in the evolution of digital technology where progress is largely determined by access to data.
First, Moore’s Law, which described the mechanism by which semiconductor capacity was rapidly expanding in the last decades of the 20th century, marched us into an age of super-fast computing hardware. Then, the arrival of the internet in the 1990s exponentially expanded human civilization’s capacity to generate and share the digital information that’s processed and transmitted by those computers. Now, at a time when AI systems are influencing virtually all aspects of our lives – to both good and bad ends – the machine-learning algorithms that drive them are becoming the primary consumers of that data. They need it to develop ever-more-powerful predictive and behavior-adjusting algorithms.
These AI systems compete in a race for the “profits” generated by their deployment, whether in the form of payoffs to corporate shareholders, greater control for government agencies, or positive public outcomes such as improvements in our collective ability to process and respond to environmental change. That competition is giving them an ever-growing, voracious appetite for data.
Also important, this is happening at the precise time that money – containing the informational traces of much of society’s person-to-person, institution-to-institution and person-to-institution interactions – is also going digital.
Monetary digitization is being driven by governments, by companies and by communities of open-source blockchain developers. Their models fall on a spectrum between ones where a state entity manages the ledger and has full, exclusive visibility on all users’ data to open protocols run by decentralized networks that integrate cryptographic tools such as zero-knowledge proofs to shield transaction histories from everyone. In the middle lie public blockchains with no privacy protection, where the transaction data is visible to all.
Read more: The Coming Privacy Wars
The data gold mine
If you're an authoritarian government with an AI system trained to undermine the interests of foreign states you regard as enemies, you will want the data from those states’ societies to be as accessible as possible. And if you’re developing a centralized form of programmable, digital money, as well as an economy-wide distributed ledger network to bring transaction-processing efficiencies to multiple industries, you’ll take a special interest in publicly available blockchain data.
That description, of course, matches China, whose central bank digital currency, the Digital Electronic Payments system, is being rolled out to hundreds of millions of users in parallel with the development of its industrial Blockchain Services Network. As these systems develop, machine-learning programs controlled, or endorsed, by the Chinese government will gain access to voluminous economic data generated by Chinese citizens.
China’s goals go beyond making its domestic economy more efficient. It aspires to lead the world in digital technology while protecting what it views as its strategic geopolitical interests. If it is to use AI to pursue those goals then the real prize lies not in data generated by the citizens of the U.S, Europe, Japan and other democratic regions.
It’s hard to access U.S. and European data protected by firewalls within centralized corporate and government systems. In contrast, open blockchains produce publicly available information. The question for policymakers is how well that transaction source data is shielded, for example with sophisticated cryptography such as Tornado Cash.
Therein lies the problem with the U.S. Treasury Department’s recent decision effectively barring Americans from transacting with Tornado Cash. In taking the unprecedented step of treating open-source software as the equivalent of a person or a company, the U.S. has stripped the Ethereum blockchain of a vital privacy option and blown open its data.
The long arm of the law is such that crypto providers everywhere responded rapidly, if reluctantly, to Treasury’s order by blocking wallets that have transacted with Tornado Cash users. In an extension of this wave of mass self-censorship, many are now also fearful of interacting with other privacy-protecting smart contracts that could be hit with similar sanctions.
But what about North Korea?
We could say we’ve entered the post-privacy era of blockchains. What’s not being discussed is that, as digital currency and blockchain systems grow, this new phase will bring informational advantages to authoritarian regimes.
That description not only fits China and Russia, but also North Korea. That’s ironic because the reason the U.S. Treasury Department gave for its sanctions against Tornado Cash was that North Korean hackers had used it to launder funds in an attack against online video game Axie Infinity.
Last week, analytics firm Chainalysis announced it helped U.S. officials recover $30 million from that attack. One might ask, then, why it was important to shut down Tornado Cash at all.
Clearly there’s a public interest in preventing such hacks. But there are all sorts of ways in which blockchain analytics and sophisticated encryption solutions can make it hard for rogue actors to get away with money while also protecting user privacy.
Whatever the solution, regulators need to understand that the price they pay for imposing broad-based constraints on privacy is nothing short of a threat to Western democracy.