Azure AI Contentsafety Py

Install

Quick Start

Non-Obvious Patterns

• Two clients: ContentSafetyClient (analyze) and BlocklistClient (blocklist management)
• Image from file: base64-encode bytes, pass via ImageData(content=base64_str)
• 8-severity mode: AnalyzeTextOptions(text=..., output_type=AnalyzeTextOutputType.EIGHT_SEVERITY_LEVELS)
• Blocklist analyze: AnalyzeTextOptions(text=..., blocklist_names=[...], halt_on_blocklist_hit=True)

Best Practices

1. Use blocklists for domain-specific terms
2. Set severity thresholds appropriate for your use case
3. Handle multiple categories — content can be harmful in multiple ways
4. Use halt_on_blocklist_hit for immediate rejection
5. Log analysis results for audit and improvement
6. Consider 8-severity mode for finer-grained control
7. Pre-moderate AI outputs before showing to users