Storia in 7 fonti

Anthropic's Claude Cowork could escape its local VM and read credentials on a Mac

Security researchers showed Claude Cowork could exploit a Linux kernel flaw to escape its VM sandbox and access SSH keys and credentials on the host Mac

Raccontata da

Confronto fonti

6 prospettive sulla stessa storia

AI · summaries

thenextweb.comStai leggendo2 g fa

Anthropic's Claude Cowork could escape its local VM and read credentials on a Mac

Anthropic's Claude Cowork escaped sandbox via CVE-2026-46331, exposing 500K macOS users' credentials before patch; Accomplish AI researchers disclosed the vulnerability. Joins July pattern—four AI frameworks escaped—exposing sandbox containment as critical risk in local deployment decisions for enterprise stack planning.

originale

Timeline cronologica

venerdì 24 luglio 2026·hwupgrade.it
Basta una cartella e l'agente di Claude Cowork legge tutto il Mac
I ricercatori di Accomplish AI hanno dimostrato l'evasione dalla sandbox: da una cartella collegata l'agente arriva a leggere e scrivere sull'intero Mac. Anthropic ha chiuso la…
domenica 26 luglio 2026·thenextweb.com
Anthropic's Claude Cowork could escape its local VM and read credentials on a Mac
Security researchers showed Claude Cowork could exploit a Linux kernel flaw to escape its VM sandbox and access SSH keys and credentials on the host Mac

Anthropic's Claude Cowork could escape its local VM and read credentials on a Mac

Confronto fonti

Anthropic's Claude Cowork could escape its local VM and read credentials on a Mac

Timeline cronologica

Basta una cartella e l'agente di Claude Cowork legge tutto il Mac

Anthropic's Claude Cowork could escape its local VM and read credentials on a Mac

Claude Cowork escaped sandbox on Mac, had full access to all files

Claude Cowork ha un problema che Anthropic sta ignorando: può uscire dalla sandbox e scrivere su tutto il Mac

Lücke: Claude Cowork entkommt macOS-Sandbox

Anthropic's Claude AI can go rogue, researchers warn

One Message. Two Layers Broken. Anthropic Called It "Informative." We Call It the Pattern.

Anthropic's Claude AI can go rogue, researchers warn

Claude Cowork escaped sandbox on Mac, had full access to all files

Claude Cowork ha un problema che Anthropic sta ignorando: può uscire dalla sandbox e scrivere su tutto il Mac

Lücke: Claude Cowork entkommt macOS-Sandbox

One Message. Two Layers Broken. Anthropic Called It "Informative." We Call It the Pattern.