As cited
Copy frozen at (site build).
ai security
Anthropic reconfigures its cool kids security program
Anthropic consolidated its Project Glasswing and Cyber Verification Program into a single three-tiered offering designed to give security teams controlled access to Claude models for vulnerability discovery and red team testing. The merged program, launched in April 2026 alongside the Mythos model, has reportedly identified over 129,000 verified vulnerabilities through partner scanning between April and July, yet fewer than 10 percent of identified critical and high-severity flaws have been patched as of the article date.
Why it matters: Security teams evaluating artificial intelligence (AI) tools for vulnerability scanning and penetration testing should understand the three access tiers (Defense, Red Team, Specialized) and associated model refusal rates, and recognize that vulnerability identification by AI models currently outpaces remediation workflows in most organizations.
- Source published
- First seen by Cybersecurity Tracker
Source attribution
Correction
Correction recorded as of .
ai security
Anthropic reconfigures its cool kids security program
Anthropic consolidated its Project Glasswing and Cyber Verification Program into a single three-tiered offering designed to give security teams controlled access to Claude models for vulnerability discovery and red team testing. The merged program, launched in April 2026 alongside the Mythos model, has reportedly identified over 129,000 verified vulnerabilities through partner scanning between April and July, yet fewer than 10 percent of identified critical and high-severity flaws have been patched as of the article date.
Why it matters: Security teams evaluating artificial intelligence (AI) tools for vulnerability scanning and penetration testing should understand the three access tiers (Defense, Red Team, Specialized) and associated model refusal rates, and recognize that vulnerability identification by AI models currently outpaces remediation workflows in most organizations.
- Source published
- First seen by Cybersecurity Tracker
Source attribution
Correction
Correction recorded as of .
ai security
Anthropic reconfigures its cool kids security program
Anthropic consolidated its Project Glasswing and Cyber Verification Program into a single three-tiered offering designed to give security teams controlled access to Claude models for vulnerability discovery and red team testing. The merged program, launched in April 2026 alongside the Mythos model, has reportedly identified over 129,000 verified vulnerabilities through partner scanning between April and July, yet fewer than 10 percent of identified critical and high-severity flaws have been patched as of the article date.
Why it matters: Security teams evaluating artificial intelligence (AI) tools for vulnerability scanning and penetration testing should understand the three access tiers (Defense, Red Team, Specialized) and associated model refusal rates, and recognize that vulnerability identification by AI models currently outpaces remediation workflows in most organizations.
- Source published
- First seen by Cybersecurity Tracker