0x233b3604233b…233b3601
AI Could Turn Crypto Security Into a 'Slaughterhouse'
Solana and Anchorage executives said AI attacks will intensify over the next 12 months as North Korean hackers target crypto firms through poisoned software, fake interviews and deepfakes.

Outputs
More than $7.5 billion in crypto was stolen this year, according to Solana Foundation CISO Michael Coates.
Most of this year's losses came from a Binance hack that drained some $6 billion; the FBI blamed North Korea, Coates said.
AI systems moved from finding flaws to carrying out full attacks on their own over the past three to six months, Mónica said.
North Korean hackers have stolen $400 million through deepfake calls alone, Coates said.
STRIDE has 18 core controls, and protocols representing more than 74% of Solana DeFi TVL have gone through it, Coates said.
More than $7.5 billion in cryptocurrency has been stolen this year, Solana Foundation chief information security officer Michael Coates said at the Breakpoint conference in London, where executives warned that autonomous AI attacks will intensify risks for crypto firms.
North Korea remains the largest threat to crypto companies, Coates said. He put cryptocurrency theft over the past decade above $13 billion and said most of this year’s losses came from last month’s Binance hack, which drained some $6 billion. Coates said the Federal Bureau of Investigation had blamed North Korea for the breach, adding that the proceeds help fund programs in North Korea.
What changes when AI can exploit alone?
Over the past three to six months, AI systems moved from finding vulnerabilities to building complete exploits and carrying out every step of an attack on their own, said Diogo Mónica, co-founder of crypto bank Anchorage Digital.
On a panel chaired by Gordon Corera, co-host of The Rest Is Classified podcast, Mónica warned: “The next 12 months are going to be absolutely brutal. They’re going to be a slaughterhouse.”
Coates said AI had dramatically lowered the cost and barrier to hacking, bringing opportunistic attention to projects that previously escaped notice. Mónica said aggressive hacking tools now let people without security knowledge simply ask a model to hack a website.
That shift turns vulnerability discovery into an automated attack process rather than an activity dependent on manual expertise, leaving crypto teams to address overlooked systems before attackers test them.
How is North Korea targeting crypto teams?
Coates described the Binance attack as a chain of familiar tactics:
- A developer downloaded poisoned software from a sponsored search result.
- An engineer later took a job interview on a work device and downloaded software as part of it.
- The foothold let attackers move through the network and reach signing services, which they used to authorize malicious transactions rather than steal the keys themselves.
Coates called fake job interviews a telltale North Korean tactic, alongside poisoned tools and deepfake video calls. He said North Korean hackers have stolen $400 million using deepfake calls alone.
The account shows how compromised employee devices can provide a path to transaction authorization without removing the cryptographic keys used to sign transactions.
When North Korean operatives cannot get hired directly, they use people in the U.S. to take jobs on their behalf, Mónica said. Those workers connect remote keyboards and mice, giving the attackers access to the company’s network. The U.S. Treasury recently sanctioned 11 people over such a North Korean IT worker scheme.
Anchorage now onboards new staff in person at its U.S. office, Mónica said. Every employee works on a locked-down Chromebook and logs into remote machines for development. He said those measures remove 90% to 95% of Anchorage’s corporate attack surface.
What defenses are crypto teams adopting?
The Solana Foundation launched STRIDE, a free security framework, in spring. It sets out 18 core controls tailored to crypto. Coates said protocols representing more than 74% of DeFi’s total value locked on Solana have gone through STRIDE, including 10 of the top 11 protocols.
The foundation also uses AI to reverse-engineer newly launched programs and find flaws before attackers do, Coates said. That work has prevented more than $5 million in potential losses over the past month, according to him.
Both executives returned to basic operational controls. Mónica warned that connecting email inboxes to AI agents invites prompt-injection attacks. Coates urged teams to protect cryptographic keys with hardware security keys for logins, properly configured multisig wallets, and plans for key rotation and key loss.
Coates said: “They’re getting the entry points through the basics. They’re pivoting around to get the keys, and that’s where they’re unleashing their fury.”
Over the next 12 months, crypto teams will need to close device, email and key-management gaps before autonomous tools turn routine mistakes into signed transactions and new access paths.
via google.com (Original)
More from Daniel Okafor
Show full bio
Correspondent covering people, launches and day-to-day practice in crypto and web3 industry at Mempool Brief.
1,514 articles