option
Home
Flash News
Content
JosephGreen
JosephGreen
April 7, 2026

Adversa AI discovered a critical vulnerability in Anthropics Claude Code AI assistant. A security parsing flaw allowed malicious commands beyond the 50th sub-command in a composite instruction to bypass all safety filters. The bug, caused by a performance trade-off, impacted over 500,000 developers. Anthropic has since released version 2.1.90 to patch the issue.

Comments (0)
0/300
OR