AI-powered attacks fuel over $600 million in crypto hacks in early 2026
Crypto platforms lost over $600 million to hacks in early 2026, with North Korean-linked attackers using AI-powered deepfakes, voice manipulation and autonomous agents to target cryptocurrency infrastructure. Security experts warn that advances in AI are enabling increasingly sophisticated attacks including social engineering, phishing and supply chain compromises.
What happened
According to CertiK, the crypto industry lost over $600 million to hacks in the first few months of 2026, with two major North Korea-linked attacks in April: the $293 million Kelp DAO exploit on LayerZero's cross-chain messaging infrastructure and the $280 million Drift Protocol exploit.
North Korean-affiliated hackers reportedly used AI in a sustained social engineering attack disclosed by crypto wallet Zerion on April 15, stealing approximately $100,000 from the company's hot wallets. A threat actor known as 'Jinkusu' was also reported in April to be selling cybercrime tools using deepfakes and voice manipulation to bypass Know Your Customer (KYC) checks at banks and crypto platforms.
Reported impact
CertiK senior blockchain investigator Natalie Newson stated that real-time deepfakes, phishing attacks, supply chain compromises and cross-chain vulnerabilities will likely be the root of some of the biggest hacks in 2026, with the average size of crypto hacks reaching $19.5 million in 2025.
Newson warned that the acceleration of AI will only worsen crypto attacks 'in some aspects,' citing the emergence of more convincing deepfakes and autonomous attack agents capable of executing exploits at machine speed.
Index analysis
CertiK identified supply chain attacks as a major threat, noting that such attacks accounted for $1.45 billion in losses in 2025 across just two incidents, including the $1.4 billion Bybit hack.
The U.S. Department of the Treasury's Office of Cybersecurity and Critical Infrastructure Protection announced on April 9 that it would begin using a threat assessment project to evaluate digital asset companies, recognizing that their infrastructure has become equal to the financial sector.
Governance controls
CertiK's Newson advised crypto users to verify the authenticity of URLs and smart contracts before interacting with them and to move idle assets off exchanges to cold wallets to prevent private key exposure.
Anthropic launched Claude Mythos, an AI model claimed to identify vulnerabilities in major operating systems, in a beta version for some tech companies to support defensive efforts.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.