Internal Anthropic research test models

Incidents implicated systems

Incident 16294 Report
Anthropic Research Model Reportedly Scanned 9,000 Targets and Compromised Real Company's Application During Evaluation

2026-07-30

During an Anthropic cybersecurity evaluation with Irregular, an internal research Claude model reportedly scanned roughly 9,000 internet targets after failing to reach its fictional target. It reportedly compromised a real company's Internet-facing application using credentials from an exposed debug page and SQL injection, then stopped after recognizing that the host was real.

Related Entities

Irregular

Incidents involved as Deployer

Incident 1629
4 Reports
Anthropic Research Model Reportedly Scanned 9,000 Targets and Compromised Real Company's Application During Evaluation

Anthropic

Incidents involved as both Developer and Deployer

Incident 1629
4 Reports
Anthropic Research Model Reportedly Scanned 9,000 Targets and Compromised Real Company's Application During Evaluation

AI evaluation organizations

Incidents involved as Deployer

Incident 1629
4 Reports
Anthropic Research Model Reportedly Scanned 9,000 Targets and Compromised Real Company's Application During Evaluation

AI agent system deployers

Incidents involved as Deployer

Incident 1629
4 Reports
Anthropic Research Model Reportedly Scanned 9,000 Targets and Compromised Real Company's Application During Evaluation

Large language model developers

Incidents involved as Developer

Incident 1629
4 Reports
Anthropic Research Model Reportedly Scanned 9,000 Targets and Compromised Real Company's Application During Evaluation

AI agent system developers

Incidents involved as Developer

Incident 1629
4 Reports
Anthropic Research Model Reportedly Scanned 9,000 Targets and Compromised Real Company's Application During Evaluation

Unidentified companies compromised during Anthropic cybersecurity evaluations disclosed July 2026

Incidents Harmed By

Incident 1629
4 Reports
Anthropic Research Model Reportedly Scanned 9,000 Targets and Compromised Real Company's Application During Evaluation

Companies

Incidents Harmed By

Incident 1629
4 Reports
Anthropic Research Model Reportedly Scanned 9,000 Targets and Compromised Real Company's Application During Evaluation

Large language models

Incidents implicated systems

Incident 1629
4 Reports
Anthropic Research Model Reportedly Scanned 9,000 Targets and Compromised Real Company's Application During Evaluation

Cybersecurity AI systems

Incidents implicated systems

Incident 1629
4 Reports
Anthropic Research Model Reportedly Scanned 9,000 Targets and Compromised Real Company's Application During Evaluation

Claude

Incidents implicated systems

Incident 1629
4 Reports
Anthropic Research Model Reportedly Scanned 9,000 Targets and Compromised Real Company's Application During Evaluation

AI agent systems

Incidents implicated systems

Incident 1629
4 Reports
Anthropic Research Model Reportedly Scanned 9,000 Targets and Compromised Real Company's Application During Evaluation