Hackers are exploiting the Claude Code vulnerability to target crypto users; security risks should be taken seriously.

MiningDisasterSurvivor · 2026-03-05T21:19:23+00:00

Recently, Anthropic's Claude Code tool was exposed to a high-risk vulnerability, allowing attackers to conduct phishing attacks on cryptocurrency users. This vulnerability enables attackers to bypass user authorization to execute commands and has been confirmed to pose an actual threat. Security experts recommend users strengthen their defenses, and the industry needs to improve security review standards for AI tools.

MiningDisasterSurvivor

2026-03-05 21:19:23

Abstract generation in progress

Recently, a serious security threat has emerged: attackers are exploiting a high-risk vulnerability in Anthropic’s Claude Code tool to carry out targeted phishing attacks against cryptocurrency users. According to security researcher Adam Chester and confirmed by the SlowMist security team, this incident has shifted from a theoretical risk to a real threat.

High-Risk Privilege Escalation Vulnerability in Anthropic Claude Code

The vulnerability (CVE-2025-64755) centers on a privilege escalation and command execution flaw, most critically allowing attackers to bypass user authorization and execute commands directly. Researcher 23pds pointed out that this is similar to a previously disclosed vulnerability in the AI code editor Cursor, indicating that such AI programming tools may face systemic security weaknesses. Proof-of-concept (PoC) code has been publicly released, significantly lowering the barrier for exploiting this vulnerability.

SlowMist Security Team Confirms Active Attacks, Cryptocurrency Users as Primary Targets

Worryingly, hackers are not just testing the waters. The SlowMist security team has confirmed through threat intelligence that attackers are actively exploiting this vulnerability via phishing to target cryptocurrency community users. Due to the high asset value, crypto users have become prime targets. Successful attacks could lead to private key leaks, theft of funds, and other serious consequences.

Security Recommendations and Industry Reflection

In response to this threat, users should take proactive protective measures: avoid using Claude Code links from unverified sources, regularly update AI tool versions, and remain cautious of any requests that ask for command execution permissions. Additionally, this incident highlights that rapidly evolving AI programming tools may lack sufficient security review, underscoring the need for the industry to establish stricter security standards and vulnerability response mechanisms.

View Original

This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.