CoinDesk
Bybit says its AI-assisted security stack blocked more than $700 million in potential losses between January 1 and June 15, 2025, processing over 100,000 alerts and stopping roughly 30,000 suspicious withdrawal requests. The exchange, still recovering from the largest crypto theft on record, claims AI found high-severity vulnerabilities at up to five times the rate of manual review and cut the gap between asset discovery and testing from about two weeks to under 24 hours. Bybit disclosed the figures in a release shared with CoinDesk and noted the numbers cannot be independently verified.
Why it matters
The release lands in a month when AI has been reshaping the defender-attacker balance across crypto, not just at well-funded exchanges. BTCPay Server earlier this month suffered an attack that drained merchant Lightning nodes; the volunteer Bitcoin Red Team has spent June pointing AI models at bitcoin codebases, filing thousands of findings across hundreds of projects and helping BTCPay patch its weakness. Dozens of crypto-aligned firms, including Coinbase, last week signed an open letter asking for access to AI models they cannot pay for. Bybit built its own stack, and its figures are the first concrete metrics a major venue has put on what that tooling actually catches.
Market impact
Beyond the $700 million figure, Bybit said its AI flagged about $212 million in funds linked to fraud and blacklisted more than 10,000 addresses. The company's head of group risk control and security, David Zong, called the cybersecurity arms race an era of minutes, while stressing that human judgment still sits at the center of critical decisions. Watch whether other venues publish comparable metrics, whether the AI-access gap between well-capitalized exchanges and small developer teams widens, and whether state-sponsored attackers can match the pace.
Frequently asked questions
-
How did AI help Bybit block $700M in losses?
Bybit says AI-assisted systems scanned 1,489 public-facing assets, processed over 100,000 alerts and blocked 30,000+ suspicious withdrawal requests between Jan 1 and June 15, 2025, protecting about 20,000 users.
-
Can Bybit's $700M figure be independently verified?
No. Bybit disclosed the figures in a release shared with CoinDesk and noted the claims cannot be independently verified. The 'potential losses' figure counts withdrawals the AI flagged and blocked, not confirmed thefts in progress.
-
What was the $1.46B Bybit hack?
In February 2025, North Korea's Lazarus Group stole roughly $1.46 billion from Bybit in what remains the largest theft in crypto history. Bybit is currently pursuing legal action against the state and the group.
-
How fast is AI changing crypto security?
Bybit says its AI cut the gap between asset discovery and vulnerability testing from about two weeks to under 24 hours, with a first review of a flagged withdrawal averaging 4.7 minutes. Its head of risk control called the cybersecurity arms race an era of minutes.
-
What is the Bitcoin Red Team?
A volunteer group that spent June pointing AI models at bitcoin codebases, filing thousands of findings across hundreds of projects. It runs on donated compute and sponsored accounts, and one of its reports helped BTCPay Server patch a vulnerability after a recent Lightning node attack.
Crypto News
WuBlockchain
TheBlock
CoinTelegraph
WatcherGuru