How many CVEs does Anthropic have?

Anthropic has 36 published CVE records since 2025, including 36 in the last two years.

How many Anthropic CVEs are in CISA KEV?

None of Anthropic's CVEs are currently listed in CISA's Known Exploited Vulnerabilities catalog.

Which Anthropic products have the most CVEs?

The Anthropic products with the most published CVEs are claude code, claude sdk for python, claude sdk for typescript, claude desktop, claude.

What are the most common weakness types in Anthropic CVEs?

Anthropic's CVEs most often map to these CWE weakness types: CWE-78 (Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection')), CWE-94 (Improper Control of Generation of Code ('Code Injection')), CWE-61 (UNIX Symbolic Link (Symlink) Following), CWE-22 (Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal')).

Are there public exploits for Anthropic vulnerabilities?

Yes — 4 of Anthropic's CVEs have a known public exploit.

How many critical Anthropic vulnerabilities are there?

Anthropic has 1 critical and 24 high-severity CVEs.

What is the average severity of Anthropic CVEs?

The average CVSS base score across Anthropic's scored CVEs is 7.0.

RadicalNotion.AI

Log in Start free

36 CVEs

Anthropic Vulnerabilities

Name: Anthropic vulnerability dataset (CVEs)
Creator: RadicalNotion.AI
License: https://creativecommons.org/licenses/by-nc/4.0/

CVE security advisories and vulnerability history for Anthropic.

36

Total CVEs

Published

0

In CISA KEV

Exploited in the wild

4

Public exploits

PoC or exploit code

7.0

Avg CVSS

2025–2026

Overview

Anthropic has 36 published CVE records since 2025, of which 0 are in CISA's Known Exploited Vulnerabilities catalog and 4 have a known public exploit. The average CVSS base score across scored CVEs is 7.0.

This page aggregates every publicly disclosed vulnerability (CVE) affecting Anthropic products, with severity breakdowns, the most-affected products, the most common weakness types, and the latest disclosures.

Security scorecard

A quick read on Anthropic's vulnerability posture, as a share of its 36 published CVEs.

Average CVSS: 7.0
Critical / high: 69%
Actively exploited (KEV): 0%
Public exploit available: 11%
Patch available: 97%

Severity and exploitation

How the CVSS severity of Anthropic's CVEs breaks down, plus how many are exploited in the wild or have public exploit code.

Critical1

High24

Medium9

Low2

In CISA’s Known Exploited Vulnerabilities catalog

0

None of Anthropic's CVEs are currently listed in CISA's KEV catalog.

Public exploits

4

4 of Anthropic's CVEs have a known public exploit available.

Most affected products

The Anthropic products with the most published CVEs. Follow any product to browse its versions and vulnerabilities.

Common weakness types

The CWE weakness categories most often found in Anthropic CVEs. Follow any weakness for its full explanation.

CNAs that assign these CVEs

The CVE Numbering Authorities that publish Anthropic CVE records.

Disclosure activity by year

How many Anthropic CVEs were published each year.

2025

12

2026

24

Latest Anthropic CVEs

The most recently disclosed vulnerabilities affecting Anthropic.

Claude Code: Sandbox Escape via Git Worktree Path Confusion Allows Unsandboxed Code Execution
CWE-22
High · CVSS 7.7
EPSS 0.6%2026-06-29
Claude Code: Insecure Temporary File in /copy Command Enables Response Disclosure and Symlink-Based File Write
CWE-59
Medium · CVSS 4.4
EPSS 0.2%2026-06-29
Anthropic Claude Desktop Cowork VM Image Contents Not Validated Before Use
CWE-353
High · CVSS 8.7
EPSS 0.1%2026-06-23
Claude Code: Out-of-Band Data Exfiltration via Pre-Approved HuggingFace Domain in WebFetch
CWE-183
Medium · CVSS 6.0
EPSS 0.5%2026-06-23
Claude Desktop: Local Privilege Escalation via Directory Junction in CoworkVMService
CWE-269
High · CVSS 8.5
EPSS 0.2%2026-05-13
Claude Desktop: SSH Host Key Verification Bypass Allows Man-in-the-Middle Attack on Remote Sessions
CWE-297
High · CVSS 7.4
EPSS 0.1%2026-05-13
Claude Code arbitrary code execution via git worktree commondir trust dialog bypass
CWE-77
High · CVSS 7.7
EPSS 0.3%2026-05-05
Claude SDK for TypeScript has Insecure Default File Permissions in Local Filesystem Memory Tool
CWE-732
Medium · CVSS 4.8
EPSS 0.1%2026-05-04
Claude Code: Sandbox Escape via Symlink Following Allows Arbitrary File Write Outside Workspace
CWE-61
High · CVSS 7.7
EPSS 0.7%2026-04-21
Claude Code: Insecure System-Wide Configuration Loading Enables Local Privilege Escalation on Windows
CWE-426
Medium · CVSS 5.4
EPSS 0.1%2026-04-17
Claude SDK for TypeScript: Memory Tool Path Validation Allows Sandbox Escape to Sibling Directories
CWE-41
Medium · CVSS 6.3
EPSS 0.3%2026-03-31
Claude SDK for Python: Insecure Default File Permissions in Local Filesystem Memory Tool
CWE-732
Medium · CVSS 4.8
EPSS 0.1%2026-03-31

Track new Anthropic CVEs as they are disclosed and get AI-written analysis and remediation guidance.

Monitor Anthropic CVEs

Other vendors

Browse vulnerabilities for other tracked vendors.

microsoft36,363 CVEs linux22,544 CVEs Google16,477 CVEs Apple15,074 CVEs Oracle12,706 CVEs Debian10,225 CVEs IBM8,823 CVEs Red Hat8,086 CVEs

Browse all vendors

Frequently asked questions

Common questions about Anthropic vulnerabilities.

How many CVEs does Anthropic have?: Anthropic has 36 published CVE records since 2025, including 36 in the last two years.
How many Anthropic CVEs are in CISA KEV?: None of Anthropic's CVEs are currently listed in CISA's Known Exploited Vulnerabilities catalog.
Which Anthropic products have the most CVEs?: The Anthropic products with the most published CVEs are claude code, claude sdk for python, claude sdk for typescript, claude desktop, claude.
What are the most common weakness types in Anthropic CVEs?: Anthropic's CVEs most often map to these CWE weakness types: CWE-78 (Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection')), CWE-94 (Improper Control of Generation of Code ('Code Injection')), CWE-61 (UNIX Symbolic Link (Symlink) Following), CWE-22 (Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal')).
Are there public exploits for Anthropic vulnerabilities?: Yes — 4 of Anthropic's CVEs have a known public exploit.
How many critical Anthropic vulnerabilities are there?: Anthropic has 1 critical and 24 high-severity CVEs.
What is the average severity of Anthropic CVEs?: The average CVSS base score across Anthropic's scored CVEs is 7.0.

References

Vulnerability data is sourced from the CVE Program; severity, KEV, and exploit signals are aggregated by RadicalNotion.AI.

Track Anthropic vulnerabilities

Monitor new Anthropic vulnerabilities as they are disclosed, with AI-written analysis and remediation guidance.

Start free See pricing