Lessons from finance in the battle against deepfakes

Stay informed with free updates

This week, a grim new record was set in the business world: fraudsters used a deepfake bot to trick an employee of a Hong Kong company into transferring $25mn. This is not the first time such a fraud has occurred. But what makes this case chilling is its unprecedented scale — and the fact that artificial intelligence was used to create a fake of the company’s chief financial officer. Yes, really.

And this is not the only current bot scare. Last autumn, Michal Šimečka, a Slovakian politician, lost an election after a fake recording of him went viral. That is ominous ahead of the 2024 US election.

And last month, a pornographic bot of the singer Taylor Swift went viral. That sparked justifiable outrage. But what is doubly horrifying is how widespread such bots have become. To take one example, human rights workers tell me that in the Middle East they are increasingly unleashed against female activists to shame them into silence.

So what can be done? Thankfully, policymakers are finally waking up. In the US, politicians such as Senator John Hickenlooper are pushing for state and federal rules around bots. This week, EU negotiators agreed on measures to criminalise the creation and dissemination of harmful bots. India is introducing “explicit” new rules, as well.

However, the EU rules only take hold in 2027, and Congress has not yet even discussed the proposed US bills. So another way to approach the problem might be not just to create new AI-linked laws, but to repurpose existing rules from other business spheres.

This is already under way with laws around privacy rights in the media, or intellectual property in creative industries and the consumer goods sector. However, another hitherto overlooked possibility is the world of finance.

The starting point is to recognise that bots are essentially “counterfeit” humans, as the philosopher Daniel Dennett has noted. This suggests, as historian and author Yuval Harari recently told me, that we should explore how governments have battled “counterfeits” elsewhere — most notably, through history, in the sphere of money.

In some senses, the lessons that history teaches us are discouraging. Consider the experience of Benjamin Franklin, one of America’s founding fathers, whose face now sits on the $100 bill. In the 18th century he pioneered measures to fight money counterfeiting, which had proliferated during the Revolutionary war. According to scientists at the University of Notre Dame, this initiative, which was at the time arguably the largest ever seen, used innovations with dye and paper.

For a while it worked. But fraudsters fought back, and by the 19th century a third to half of all American bills in circulation were fake, according to Congress. (Which is presumably why Franklin is more famous today for his innovations with bifocals and lightning rods.)

Yet here is the striking, and encouraging, twist: although digitisation now makes it cheaper than ever to create counterfeits, less than 0.01 per cent of all bills in circulation globally are fake, according to the US Fed. For sterling, it is 0.0031 per cent.

Why has fakery fallen? The Boston Fed attributes this to innovative detection systems that increasingly employ AI, as well as to rising inter-institutional and cross-border co-operation, via Interpol and other bodies. However, I suspect another crucial issue is that the public knows fake money is illegal, since the laws are clear, and barely a week goes by in America without news of successful prosecutions.

Mostly, these prosecutions target the counterfeit creators themselves, who can face up to 15 years in jail and a $15,000 fine. But if people knowingly use counterfeits to deceive others they can also face fraud charges. And while institutions that unwittingly handle counterfeits have hitherto largely avoided prosecution, this could still change.

In the EU, Amazon is being sued over the sale of counterfeit goods on its platform. If that case succeeds, it could set a precedent that might spread. Either way, banks clearly face strong incentives to do extensive monitoring to prevent counterfeits.

Of course, bots are not the same as bank bills: they go viral faster than notes, are easier to produce and tend to harm individuals, not institutions such as central banks. And artists might argue that the right to freely create bots is part of free expression.

But the key takeaway from financial history is that it is possible to create a cultural frame that criminalises counterfeiting. Now that would certainly not address all the deepfake problems online. But it could help to deter some of them in a crucial election year.

And looking at the world of money underscores another point, as Harari notes. AI need not always be viewed and treated as if it sits in a class of its own, whatever the investor hype and political fear. Fakery can occur with many different technologies.

So let’s hope that Congress will get anti-bot laws passed, and the EU accelerates its rules. But let us also hope that tech companies and policymakers take a lesson from the world of money. Perhaps we should all revisit Franklin, and the tale of the 18th-century ink.

Articles You May Like

Investors sit back ahead of primary slate, FOMC minutes
Israel’s ‘don’t reward them’ argument misses a moment of opportunity
2 out of 5 industrial stocks are at record highs. Here’s our post-earnings outlook on all of them
Judge expected to issue ruling in Trump business fraud trial Friday
English private schools focus on recruiting overseas as Labour VAT hit looms