There is a change coming to crypto crime, and North Korea’s state-backed hackers are within the vanguard.
There isn’t any longer any want for dozens of expensively educated programmers to investigate blockchain code and good contracts for vulnerabilities, it is now doable to set AI to the duty, in line with Kostas Kryptos Chalkias, co-founder and chief cryptographer of Mysten Labs.
Giant language fashions characterize a larger risk to the business than quantum computing, which probably would work so quick that the encryption algorithms used develop into out of date. Pyongyang’s cyber models, liable for stealing an estimated $2 billion in crypto already this yr, have begun integrating massive language fashions into almost each stage of their assaults: reconnaissance, phishing, code evaluation and laundering the proceeds, he stated.
“AI is one of the best device I’ve ever had as a white-hat hacker,” Chalkias stated in an interview with CoinDesk. “And you may think about what occurs when it’s within the incorrect fingers.”
AI-driven theft at file scale
The Lazarus Group, the nation’s most infamous hacking unit, has already set information in 2025. Investigators say the $1.5 billion Bybit breach in February, attributed by the FBI to North Korean operatives, was the biggest crypto hack in historical past.
What’s new this yr, Chalkias stated, is automation. Utilizing AI fashions much like ChatGPT and Claude, attackers can now analyze open-source codebases throughout a number of blockchains, flag possible vulnerabilities and mirror profitable exploits from one ecosystem to a different.
“AI can mix information from earlier hacks and instantly spot the identical weak spot elsewhere,” he defined. “A human can’t manually scan hundreds of good contracts, however an AI can do it in minutes.”
That means turns a small cell of state hackers into one thing resembling a digital industrial complicated. “You’ll be able to scale your assault floor with a single immediate,” Chalkias stated. “That’s what makes it harmful.”
Safety researchers at Microsoft and Mandiant have labored collectively on the pattern, documenting an increase in AI-assisted phishing, deepfake impersonations and artificial job purposes utilized by North Korean operatives posing as Western software program builders.
The regime’s AI toolkit now spans the complete intrusion chain from social engineering, code evaluation and cross-chain exploitation to laundering, which makes use of pattern-recognition algorithms to trace liquidity paths via mixers and OTC brokers, automating obfuscation.
Quantum: Nonetheless distant, however looming
For years, the business’s doomsday state of affairs centered on quantum computing: Machines highly effective sufficient to crack bitcoin’s SHA-56 encryption and unlock tens of millions of dormant cash.
Chalkias, who holds a doctorate in identity-based cryptography and has spent greater than a decade researching post-quantum algorithms, stays calm.
“There’s no proof as we speak that any laptop, even a labeled one, can break fashionable cryptography,” he stated. “We’re not less than 10 years away from that.”
He credit organizations just like the U.S.’ Nationwide Safety Company and Enisa, the European Union’s company for cybersecurity, for pushing early adoption of quantum-safe requirements, and frames these efforts as preventive quite than reactive.
Mysten Labs, developer of the Sui blockchain, is already constructing migration instruments that can let customers shift funds into quantum-resistant accounts when the time comes. Chalkias worries that AI would possibly convey that date nearer by serving to physicists design new supplies or error-correction strategies.
“The mix of AI and quantum is what freaks me out,” he stated. “We would have created a brand new species, and we will’t predict its tempo.”
The larger and quicker risk
Whereas quantum threats stay theoretical, AI is at present breaking issues at a charge of knots.
DeFi platforms are notably uncovered, Chalkias stated, as a result of open-source code permits AI fashions, pleasant or hostile, to comb via each line of logic.
“AI makes it trivial to seek out mirrored bugs throughout protocols,” he stated. “If one oracle fails, dozens could share the identical flaw.”
He predicts that regulators will quickly require steady, AI-aware auditing for exchanges and smart-contract platforms, primarily a standing red-team that reruns vulnerability scans each time a serious AI mannequin is up to date.
“Every new model of GPT or Claude finds completely different weaknesses,” he stated. “In case you’re not testing towards them, you’re already behind.”
Nonetheless, AI is a double-edged sword and can be utilized defensively in addition to in assault.
Meaning embedding AI-based safety into wallets, custodians, and exchanges, and re-auditing good contracts repeatedly. It additionally means getting ready for the long-term quantum transition now, earlier than regulation forces it.
“Until we construct anti-AI defenses into the whole lot we do,” he warned, “we’ll all the time be one step behind.”
North Korea’s Subsequent Transfer
Past pure hacking, North Korea has begun experimenting with AI-generated propaganda and disinformation, in line with Western intelligence companies. However Chalkias stated he believes the nation’s most potent near-term weapon stays AI-enhanced social engineering.
When requested whether or not North Korea may ever construct the primary quantum laptop, he laughed.
“No,” he stated. “The true race is between the U.S. and China. North Korea will overuse AI for phishing, deepfakes and deception. That’s the place their power lies.”
Even with out quantum functionality, AI lets hackers simulate legit customers, mimic transactions, and launder funds with unprecedented subtlety.
“They don’t want quantum to interrupt crypto,” Chalkias stated. “They simply want AI to make the assault invisible.”