Across Protocol warns users to avoid app amid potential crypto security incident
Crosschain bridge operator Across Protocol is telling users to stay away from its application and avoid connecting wallets as it investigates reports of drained funds. The warning adds to a fresh round of security concerns in crypto infrastructure, with users on social platform X claiming losses after using the service.
Highlights
- Across Protocol urges users to avoid its app and halt all transactions amid investigation of a potential security breach reported on June 11.
- Multiple Across users on X report wallet drains, including a $9,000 USDC loss and loss of 7,053 USDT0, with allegations of website compromise.
- NEAR AI discloses attackers exploited a SGLang vulnerability, exposing account details of 100 users and 18 AI agents, patched and remediated by June 10.
Security warning and user reports
As reported by Cointelegraph, citing Across Protocol in a post on X on Tuesday, the company advises users not to access its app, connect wallets, or sign transactions and approvals until further notice while it investigates a potential security incident.Across, which enables transfers of digital assets between Ethereum and other blockchain networks, does not specify the nature of the incident or say whether its website is compromised. Several users on X say they lost funds, including one who claims a loss of $9,000 in USDC and another who reports losing 7,053 USDT0 in seven transactions on the Plasma blockchain. A third user says their wallet is drained after using the bridge multiple times and alleges the website is compromised.
Across asks affected users to share transaction details as it reviews the reports. Cointelegraph says it has not independently verified the claims, and the protocol does not immediately respond to a request for further comment.
Broader pressure on crypto security
In a separate incident, NEAR AI discloses on Monday that attackers exploit a vulnerability in its AI infrastructure, exposing account details for about 100 users and data from 18 AI agents.According to a Nov. 9 incident report, the attackers use a remote code execution vulnerability in SGLang, an open-source AI inference engine, to access an internal server and other infrastructure. The exposed information includes user names, email addresses, agent chat histories, settings and logs, while two customers have API keys exposed. NEAR AI says attackers may also observe requests processed on the compromised server, including some running in confidential mode.
The company says it patches the vulnerability, revokes exposed credentials and restores access to its cloud services. The two disclosures highlight continuing operational and reputational risks for crypto and adjacent blockchain platforms, where security incidents can quickly affect user confidence and transaction activity.
We have also analyzed Cohere’s planned AISLE acquisition to expand AI-powered software vulnerability detection for enterprise security.