
short
- Ostium lost about US$18 million in an oracle exploit on Wednesday.
- The attackers used a compromised Oracle signing key to provide fake price reports with a future date.
- This vulnerability drained nearly a third of the protocol’s liquidity.
Ostium lost nearly $18 million on Wednesday after attackers compromised its oracle signing key and manipulated the perpetual decentralized exchange’s price feed to make fake trading profits, according to security firm Blockaid.
In a mail In Stable coin USDC – from Ostium’s liquidity vault.
“We are aware of the issue with the OLP vault,” Ostium said books On X. “We have paused all trading. The team is investigating.”
Built on resolution, The door Linked perpetual futures offers Real world origins Including stocks, commodities, foreign exchange markets and indices. It operates as a decentralized exchange, or Direct implementationwhich means that users remain largely in control of their funds and do not provide personally identifiable information. At the time of the attack, the protocol held a total value of about $63 million, meaning that the exploit drained nearly a third of its liquidity.
The attack comes amid one of the worst years on record for DeFi exploitation. Decentralized financeDecentralized finance, or decentralized finance, refers to financial applications that run natively on blockchain networks, without external intermediaries such as banks. more than $840 million They were stolen from DeFi protocols in the first five months of 2026, incl $292 million Stolen from KelpDAO $285 million From drift protocol. Hackers also targeted Resolv Labs in June, stealing data 25 million dollars.
Security experts warn against this Progress In the field of artificial intelligence, exploit detection is accelerating.
“AI is much better at reviewing code than most people and finding potential vulnerabilities in it,” Danny Jenkins, CEO and co-founder of ThreatLockerIt was said previously Decryption. Existing AI systems are already accelerating vulnerability discovery, while newer models like Mythos can significantly expand these capabilities, Jenkins said, calling it an imminent “big problem.”
“It will only be a matter of time until someone bad gets access to it,” he said.
In May, security researcher Taylor Hornby used Anthropic’s Claude book Opus 4.8 To identify a four-year-old fake They are weak at Zcash, underscoring how leading AI models are becoming increasingly effective at finding flaws in complex software.
Daily debriefing Newsletter
Start each day with the latest news, plus original features, podcasts, videos and more.





