There is no evidence that artificial intelligence was used in the massive hack on Binance, but analysts warn that using the technology for such attacks could make them far more damaging.
Estimates for the security breach on Thursday project that about $6.3 billion in assets were affected. About $3 billion has been frozen, including more than $2.3 billion in stablecoins. Binance co-chief executive Yi He said the exchange would cover all losses. Deposits and withdrawals remain suspended.
Binance, the world’s biggest crypto exchange, said preliminary investigations suggested the attack originated from a laptop belonging to one of its engineers, which allowed the hacker to make unauthorised withdrawals.
This indicates the weakness lay not in the blockchain itself, but in how the assets were held, particularly in the exchange’s crypto wallets.
“A blockchain can be perfectly secure while the people and institutions operating around it are not. Banks have understood [securing their vaults] for centuries,” Bhaskar Dasgupta, chairman and founding board member of the Middle East Stablecoin Association, told The National. “You can build the strongest vault in the world, but you still need to worry about who has the keys.”
While there is no proof that AI was used in the hack, the technology could make crypto attacks “more interesting, and potentially more dangerous”, he said.
“It is already making social engineering, phishing, vulnerability discovery and automated attacks easier to scale. Equally, defenders are using AI,” Mr Dasgupta said. “I wouldn’t blame AI for this particular incident without evidence, but I would be surprised if it didn’t become an increasingly important part of both sides of the cybersecurity equation.”
Binance uses hot wallets, which are connected to the internet. Cold wallets, by contrast, are kept offline, storing private keys completely off-grid.
By using hot wallets, Binance trades some security for speed to give customers the experience they expect, said Goncalo Magalhaes, head of security at onchain security platform Immunefi. While no keys were leaked, the company’s own systems signed the withdrawals, he noted.
“Once software with signing permissions is compromised, it doesn’t really matter if the key itself is secure,” he told The National. “There’s no sign AI was used here ... but AI has indeed made finding weaknesses cheap, while fixing them still runs at human speed.”
Cryptocurrency platforms are not easy to infiltrate, given the multiple layers of security protecting billions of dollars’ worth of digital assets. But the threats are growing as attackers develop more sophisticated methods.
“Blockchain technology relies on advanced cryptographic security mechanisms, but vulnerabilities can still emerge in digital wallets, operating systems, access management and exchanges' underlying technical infrastructure,” Rania Gule, senior market analyst at XS.com for the Middle East and North Africa, told The National.
“This highlights a fundamental distinction: the technology's security does not automatically guarantee the security of the institutions operating it.”
AI adds a new dimension to cyber security risks by enabling attackers to identify vulnerabilities faster, automate stages of cyber attacks, analyse systems and develop increasingly sophisticated phishing and fraud techniques, she said.
The same technology, however, can also strengthen defences through more effective monitoring, detection of unusual transactions and suspicious behaviour, and faster, more proactive responses to potential threats, Ms Gule added.
“We are witnessing an ongoing technological race between offensive and defensive capabilities."
