PrivacyMedium
Anthropic's Claude Code AI Coding Tool Reportedly Used Hidden Unicode Markers to Classify Custom API Routing
- Occurred
- Apr 2, 2026
- Published
- Jun 30, 2026
- Source
- AI Incident Database
Summary
Researchers found that Anthropic's Claude Code AI coding tool classified custom API endpoint hostnames and two system timezones, encoding the results in system prompts with visually similar Unicode apostrophes and altered date separators. Anthropic said the mechanism was an anti-reseller and anti-distillation experiment launched in March and removed it after disclosure.
Developers
- Anthropic
- AI software developers
Deployers
- Anthropic
- AI service providers
Harmed parties
- AI coding assistant users
- Privacy
- Claude Code users
Classification
Category confidence: 85%
Severity confidence: 80%
Rationale: User discovered steganographic marking in AI code tool, raising privacy concerns about unauthorized data tracking.
Classifier: Qwen/Qwen3.5-35B-A3B-FP8