Cathedral Used AI Agents to Access Anthropic Internal Business Systems
AI agents
Security firm Cathedral leveraged AI agents to exploit third‑party software and gain limited access to Anthropic’s internal business systems during an authorized test. Anthropic confirmed the breach, blocked the access, and noted incomplete adherence to testing rules.
why this severityAttackers accessed internal business systems, confirming exploitation.
- first report
- 26 Oct 2026
- last update
- 26 Oct 2026
What to do
Review and strengthen internal access controls and monitor AI agent usage.
Timeline · 1
Sources · 1
Research and catalogues
- 2026-10-26Cybersecurity Firm Cathedral Reportedly Used AI Agents to Access Anthropic's Internal Business SystemsAI Incident Database · incidentdatabase.ai/cite/1789
Titles and links belong to their publishers. The summary above is written by gcdRadar in its own words; data is CC BY-SA 4.0.
Classification
- AI role
- AI-enabled attack
- categories
- AI-assisted intrusion
- severity
- 3 · Confirmed exploitation
- targets
- Anthropic
- owasp llm top 10
- no entry for these categories
- mitre atlas
- no technique for these categories
- dates
- happened 26 Oct 2026 · first reported 26 Oct 2026
- origin
- AI Incident DatabaseDerived from the AI Incident Database (CC BY-SA 4.0).
- record
- confidence 0.88 · created 2026-11-05 18:55Z · updated 2026-11-10 00:11Z
Lane, AI role, categories and severity come from the classifier; the OWASP and ATLAS rows are derived from the categories. How these are assigned
Copy a brief
This record as a short Markdown brief, for a ticket, a chat or a coding agent.
the brief as text
# Cathedral Used AI Agents to Access Anthropic Internal Business Systems https://gcdradar.vercel.app/incidents/cathedral-used-ai-agents-to-access-anthropic-internal-busine · AIR-2026-1784 · closed 2026-11-09 - Lane: Breaches & attacks - Severity: 3 (Confirmed exploitation): Attackers accessed internal business systems, confirming exploitation. - Fix status: No fix needed - First reported: 2026-10-26 - AI role: AI-enabled attack - Products: AI agents ## Summary Security firm Cathedral leveraged AI agents to exploit third‑party software and gain limited access to Anthropic’s internal business systems during an authorized test. Anthropic confirmed the breach, blocked the access, and noted incomplete adherence to testing rules. ## What to do Review and strengthen internal access controls and monitor AI agent usage. ## Latest updates - 2026-10-26: Cybersecurity Firm Cathedral Reportedly Used AI Agents to Access Anthropic's Internal Business Systems (https://incidentdatabase.ai/cite/1789) ## Sources - 2026-10-26 AI Incident Database: Cybersecurity Firm Cathedral Reportedly Used AI Agents to Access Anthropic's Internal Business Systems https://incidentdatabase.ai/cite/1789 Verify against the vendor's advisory before changing production. Data CC BY-SA 4.0, https://gcdradar.vercel.app.
Related
- closedOpenAI bots Attacker Scans for Credentials by Impersonating AI Bots
Breaches & attacks · Exploited · since 28 Oct 2026
AI - closedChrome Anthropic AI Model Chains Flaws Into Working Exploits in Lab Research
Research · since 28 Oct 2026
AI - closedClaude Anthropic demonstrates AI-assisted vulnerability discovery using Claude
Research · since 27 Oct 2026 · updated 29 Oct 2026
AI - closedClaude Claude Demonstrates Multistage Cyber Attacks on Realistic Ranges
Research · since 20 Oct 2026
AI - closedClaude Mythos 5 AISI Evaluation Agents Conduct Unsanctioned Real-World Cyber Activity
Research · since 25 Jul 2026
AI - closedClaude Opus 4.7 Claude Opus 4.7 Compromises Real Production Infrastructure in Evaluation
Research · since 30 Jul 2026
AI