§Search45 dispatches

“Anthropic”

45 stories match, newest first.

Friday, July 10

AI Coding Agents Vulnerable to Exploitation by Malicious Code
Premium
AI Now Institute's 'Friendly Fire' shows autonomous AI coding agents can be tricked into executing attacker code during self-approved scans.
Jul 10·thehackernews.com
cve & vulnerabilities ai & llm security thehackernews.com
GhostApproval Flaw in AI Coding Assistants Poses Security Risks
Premium
Wiz-discovered GhostApproval symlink flaw lets six AI coding assistants run attacker-controlled code by tricking file approvals.
Jul 10·thehackernews.com
cve & vulnerabilities ai & llm security thehackernews.com

Friday, June 26

Curl Update Addresses 25-Year-Old Security Vulnerability
Premium
Curl patches a 25-year-old vulnerability, fixing 18 CVEs in a record release; update all curl-enabled products.
Jun 26·securityweek.com
cve & vulnerabilities securityweek.com

Monday, June 15

Anthropic Defends Claude Fable 5 Against Jailbreak Allegations
Premium
Industry debate on Fable 5 jailbreak claims; consider model risk management and vendor assurances.
Jun 15·securityweek.com
ai & llm security securityweek.com

Wednesday, April 29

Critical SQL Injection Vulnerability in LiteLLM Exposes Cloud Credentials
Critical pre-auth SQL injection in LiteLLM openly exploited in the wild, exposing cloud credentials.
Apr 29·cybersecuritynews.com
cve & vulnerabilities cloud security ai & llm security cybersecuritynews.com

Get these articles delivered to your inbox.