Across Protocol warns users to avoid app amid reports of drained wallets
Quick Look
- Crosschain bridging protocol Across warned users Tuesday not to access its app or connect wallets while investigating a potential security incident, as users reported losses including $9,000 in USDC and 7,053 USDT0.
- Separately, NEAR AI disclosed a Monday breach exploiting an SGLang vulnerability that exposed data from approximately 100 users and 18 AI agents.
AI-generated summary
Why It Matters
Across is a crosschain bridging protocol that allows users to transfer digital assets between Ethereum and other blockchain networks. NEAR AI disclosed a separate security incident involving a vulnerability in its AI infrastructure.
Crosschain bridging protocol Across, which allows users to transfer digital assets between Ethereum and other blockchain networks, warned users Tuesday not to access its app or connect their wallets while it investigates a potential security incident.
In a Tuesday X post, Across advised users against signing transactions or approvals until further notice, without specifying the nature of the incident or whether its website had been compromised.
On X, one user claimed to have lost $9,000 in USDC, while another reported losing 7,053 USDT0 across seven transactions on the Plasma blockchain.
A third user said their wallet was drained after using the bridge several times, alleging that the website had been compromised. Across asked the users for transaction details to investigate their reports.
Cointelegraph has not independently verified the reports. Across Protocol did not immediately respond to Cointelegraph’s request for more details.
Across Protocol responds to user reporting $9,000 USDC loss. Source: X user bilm01
NEAR AI discloses separate security breach
In a separate security incident, NEAR AI disclosed Monday that attackers exploited a vulnerability in its AI infrastructure, exposing account details belonging to approximately 100 users and data from 18 AI agents.
According to a Nov. 9 incident report, attackers exploited a remote code execution vulnerability in SGLang, an open-source AI inference engine, to gain access to an internal server and other infrastructure. The compromised data included user names, email addresses, agent chat histories, settings and logs, while two customers had API keys exposed.
NEAR AI said attackers may also have observed requests processed on the compromised server, including those running in confidential mode. The company said it had patched the vulnerability, revoked exposed credentials and restored access to its cloud services.
What to Watch
AI outlook — possibilities, not facts
Across Protocol will release further details about the security incident after completing its investigation
Likely · Within days
NEAR AI will monitor for additional unauthorized access following its patch and credential revocation
Likely · Within days
Open Questions
- What is the nature of the Across security incident?
- Was the Across website compromised?
- Have the user loss claims been verified?
- How many users were affected by the NEAR AI breach?
