agents
latest
false
- Automation Cloud と Test Cloud
- Automation Cloud (専有型) および Test Cloud (専有型)
重要 :
このコンテンツの一部は機械翻訳によって処理されており、完全な翻訳を保証するものではありません。
新しいコンテンツの翻訳は、およそ 1 ～ 2 週間で公開されます。
Agents リリース ノート
最終更新日時 2026年5月21日
May 21, 2026
New out-of-the-box guardrails
Three new out-of-the-box guardrail types are now available in public preview. You can enable them directly from your agent's settings in Studio Web, alongside the existing PII detection and Prompt injection guardrails.
- Harmful content – Detects hate speech, self-harm, sexual content, and violence in LLM interactions. Applies pre- and post-execution to the selected scope: agent, LLM, or tools. Powered by Azure AI Content Safety.
- IP protection – Detects intellectual property leakage in model-generated text and code. Applies post-execution only to the selected scope: agent or LLM. Powered by Azure AI Content Safety.
- User prompt attacks – Detects prompt injection and jailbreak attempts before they reach the LLM. Applies pre-execution only. Powered by Microsoft Content Safety.
For configuration details, see Out-of-the-box guardrails.