> TritonInfoSec News
Home / Jul 25, 2026 / Story
0
#8 The Hacker News general July 23, 2026 at 13:27 UTC

Claude Cowork Flaw Could Let AI Agent Escape Its VM and Access Mac Files

By [email protected] (The Hacker News)

AI Summary

Accomplish AI researchers discovered a sandbox escape vulnerability in Anthropic's Claude Cowork product that allows a malicious or manipulated AI agent to break out of its Linux VM and read or write files anywhere on the host macOS filesystem. The vulnerability affects approximately 500,000 macOS users running the Claude Cowork application. The flaw illustrates a systemic risk in AI agent sandboxing architectures where VM isolation boundaries are insufficiently enforced.

Read full article → https://thehackernews.com/2026/07/claude-cowork-fl…

Relevance score: 76.0/100

# More from July 25

  1. 1
    Russian Espionage Group Exploited Zimbra Zero-Day to Steal Mail and 2FA Codes The Hacker News
  2. 2
    Certighost Exploit Lets Low-Privileged Active Directory Users Impersonate a Domain Controller The Hacker News
  3. 3
    Bing Images Flaws Let Crafted SVGs Run Commands as SYSTEM on Microsoft's Servers The Hacker News
  4. 4
    Kimi K3 Agents Found Redis Zero-Days and Built RCE Exploit, Researchers Say The Hacker News
  5. 5
    Hacker Runs Hermes AI Agent Unattended for Post-Exploitation at Thai Finance Ministry The Hacker News
  6. 6
    Clop ransomware targets Windchill, FlexPLM in data theft attacks BleepingComputer
  7. 7
    Check Point warns of SmartConsole zero-day exploited in attacks BleepingComputer
  8. 9
    New RefluXFS Linux flaw lets attackers gain root privileges BleepingComputer
  9. 10
    ChatGPT AgentForger Flaw Could Deploy Rogue Workspace Agents via a Phishing Link The Hacker News