# Hackers are stealing Claude tokens from subscribers

Published: 2026-09-08 · Updated: 2026-09-08 · Category: Safety

Source page: https://www.sipply.dev/story/hackers-are-stealing-claude-tokens-from-subscribers-b9f522ad · 中文: https://www.sipply.dev/zh/story/hackers-are-stealing-claude-tokens-from-subscribers-b9f522ad

## Summary

Recent reports indicate that hackers have been targeting users of Anthropic's AI model, Claude, by stealing tokens associated with their accounts. This unauthorized access has resulted in unexpected token consumption and financial losses for affected users. One notable case involved a user named Grant De Swardt, who found that his session key had been compromised, leading to the suspension of his account and a subsequent refund from Anthropic. The incident highlights vulnerabilities in user account security and the potential for financial repercussions stemming from such breaches.

## Key takeaways

- Unauthorized access to Claude tokens is leading to financial losses for users.
- Compromised session keys are a significant security risk for AI service subscribers.
- Companies like Anthropic are taking steps to address these security breaches by suspending accounts and issuing refunds.

## Why it matters

This incident underscores the importance of robust security measures in protecting user accounts and preventing unauthorized access to digital services, particularly in the AI sector.

- **Product managers:** Product managers should prioritize security features in their product roadmaps to protect users and maintain trust in their services.
- **Developers:** Developers need to implement strong authentication and session management practices to safeguard against token theft and other security vulnerabilities.
- **Students and job seekers:** For students and job-seekers, understanding the implications of cybersecurity in AI services is crucial, as it highlights the importance of security skills in the tech job market.

## Sources

- [Hackers are stealing Claude tokens from subscribers](https://techcrunch.com/2026/09/08/hackers-are-stealing-claude-tokens-from-subscribers) (TechCrunch AI, 2026-09-08)

## 中文

### 黑客正在从订阅者那里窃取Claude令牌

最近的报告显示，黑客针对Anthropic的AI模型Claude的用户，通过窃取与其账户相关的令牌进行攻击。这种未经授权的访问导致受影响用户的令牌消耗异常和财务损失。一个显著的案例涉及用户Grant De Swardt，他发现自己的会话密钥被泄露，导致账户被暂停，并从Anthropic获得退款。该事件突显了用户账户安全的脆弱性以及此类泄露可能带来的财务后果。

**要点：**

- 对Claude令牌的未经授权访问导致用户财务损失。
- 被泄露的会话密钥对AI服务订阅者构成重大安全风险。
- 像Anthropic这样的公司正在采取措施，通过暂停账户和发放退款来应对这些安全漏洞。

**为什么重要：** 这一事件强调了在保护用户账户和防止未经授权访问数字服务方面，尤其是在AI领域，强大安全措施的重要性。

---
Sipply: Sip AI News Simply. https://www.sipply.dev