---
title: "ai_governance.prompt.flagged"
method: POST
path: "ai_governance.prompt.flagged"
tags: ["Webhooks"]
---

# ai_governance.prompt.flagged

`POST ai_governance.prompt.flagged` (webhook)

A prompt a user submitted to an AI product was flagged as risky.

**Subscription filters**

- `product` — one of: `ABACUS_CHAT`, `AMPCODE`, `ANAKIN_CHAT`, `ANDI_CHAT`, `BIG_AGI_CHAT`, `CANVA_CHAT`, `CHATBOX_AI_CHAT`, `CHATGLM_CHAT`, `CHATGOT_CHAT`, `CHATGPT_AGENTS`, `CHATGPT_CHAT`, `CHATGPT_CODEX`, `CHATHUB_CHAT`, `CHATON_CHAT`, `CHATPLAYGROUND_CHAT`, `CHERRY_STUDIO_CHAT`, `CLAUDE_CHAT`, `CLAUDE_CODE`, `CLAUDE_COWORK`, `CLAUDE_DESIGN`, `COPILOT_CHAT`, `CURSOR_AGENTS`, `CURSOR_CHAT`, `DEEPSEEK_CHAT`, `DOUBAO_CHAT`, `DUCK_AI_CHAT`, `ERNIE_CHAT`, `FANAR_CHAT`, `FELO_CHAT`, `FIREWORKS_CHAT`, `GEMINI_CHAT`, `GIGACHAT_CHAT`, `GLOBALGPT_CHAT`, `GOVAI_CHAT`, `GRAMMARLY_CHAT`, `GROK_CHAT`, `IASK_CHAT`, `IFLYTEK_SPARK_CHAT`, `INNER_AI_CHAT`, `JAIS_CHAT`, `KIMI_CHAT`, `KIRO_CLI`, `KIRO_CREW`, `KIRO_WEB`, `KOMO_CHAT`, `LMARENA_CHAT`, `LOVABLE_CHAT`, `LUZIA_CHAT`, `M365_COPILOT_CHAT`, `M365_COPILOT_EMBEDDED`, `MAGAI_CHAT`, `MAGICA_CHAT`, `MERLIN_CHAT`, `META_AI_CHAT`, `MISTRAL_CHAT`, `MONDAY_CHAT`, `MONICA_CHAT`, `MORPHIC_CHAT`, `MUSE_AI_CHAT`, `NANO_GPT_CHAT`, `NOTION_CHAT`, `ONE_MIN_AI_CHAT`, `OPENROUTER_CHAT`, `PERPLEXITY_CHAT`, `PERPLEXITY_COMPUTER`, `PI_CHAT`, `POE_CHAT`, `QOLABA_CHAT`, `QWEN_CHAT`, `REPLIT_CHAT`, `RIPPLING_CHAT`, `SENSECHAT_CHAT`, `SHAKUDO_CHAT`, `SIDER_CHAT`, `STEPFUN_CHAT`, `T3_CHAT`, `TYPINGMIND_CHAT`, `USE_AI_CHAT`, `VENICE_CHAT`, `WRTN_CHAT`, `YANDEX_ALICE_CHAT`, `YOU_COM_CHAT`, `YUANBAO_CHAT`, `Z_AI_CHAT`

## Payload

- WebhookEnvelopeAiGovernancePromptFlaggedPayload: The CloudEvents envelope wrapping every webhook event
  - `data` WebhookEnvelopeDataAiGovernancePromptFlaggedPayload: The data member of the CloudEvents envelope: the event payload plus the subscription's custom data
    - `customdata` unknown
    - `payload` AiGovernancePromptFlaggedPayload: A prompt a user submitted to an AI product was flagged as risky.
      - `account_info` AiAccountInfo: The account signed in to the AI product when the prompt was submitted.
        - `logged_in_email` string: Email address of the account signed in to the AI product.
        - `plan_type` string: The AI product's plan for the signed-in account, as reported by the vendor (for example Free, Plus, Team, or Enterprise).
      - `file_names` string[]: Names of the files attached to the prompt. Empty when no files were attached.
      - `flagged_at` string, date-time: When the prompt was flagged (ISO-8601).
      - `product` 'CHATGPT_CODEX' | 'CHATGPT_CHAT' | 'CHATGPT_AGENTS' | 'CLAUDE_CHAT' | 'CLAUDE_COWORK' | 'CLAUDE_CODE' | 'CLAUDE_DESIGN' | 'COPILOT_CHAT' | 'CURSOR_CHAT' | 'CURSOR_AGENTS' | 'GEMINI_CHAT' | 'KIRO_WEB' | 'KIRO_CLI' | 'KIRO_CREW' | 'AMPCODE' | 'M365_COPILOT_CHAT' | 'M365_COPILOT_EMBEDDED' | 'PERPLEXITY_CHAT' | 'PERPLEXITY_COMPUTER' | 'DEEPSEEK_CHAT' | 'GROK_CHAT' | 'QWEN_CHAT' | 'ABACUS_CHAT' | 'CANVA_CHAT' | 'LOVABLE_CHAT' | 'REPLIT_CHAT' | 'GRAMMARLY_CHAT' | 'NOTION_CHAT' | 'RIPPLING_CHAT' | 'USE_AI_CHAT' | 'GOVAI_CHAT' | 'SHAKUDO_CHAT' | 'MONDAY_CHAT' | 'OPENROUTER_CHAT' | 'META_AI_CHAT' | 'DUCK_AI_CHAT' | 'DOUBAO_CHAT' | 'NANO_GPT_CHAT' | 'VENICE_CHAT' | 'FIREWORKS_CHAT' | 'Z_AI_CHAT' | 'MISTRAL_CHAT' | 'POE_CHAT' | 'LMARENA_CHAT' | 'MONICA_CHAT' | 'CHATON_CHAT' | 'GLOBALGPT_CHAT' | 'YOU_COM_CHAT' | 'ONE_MIN_AI_CHAT' | 'ANAKIN_CHAT' | 'ANDI_CHAT' | 'BIG_AGI_CHAT' | 'CHATBOX_AI_CHAT' | 'CHATGLM_CHAT' | 'CHATGOT_CHAT' | 'CHATHUB_CHAT' | 'CHATPLAYGROUND_CHAT' | 'FANAR_CHAT' | 'FELO_CHAT' | 'MERLIN_CHAT' | 'GIGACHAT_CHAT' | 'IASK_CHAT' | 'INNER_AI_CHAT' | 'KIMI_CHAT' | 'KOMO_CHAT' | 'LUZIA_CHAT' | 'MAGAI_CHAT' | 'MAGICA_CHAT' | 'MORPHIC_CHAT' | 'PI_CHAT' | 'QOLABA_CHAT' | 'SIDER_CHAT' | 'T3_CHAT' | 'TYPINGMIND_CHAT' | 'YANDEX_ALICE_CHAT' | 'SENSECHAT_CHAT' | 'IFLYTEK_SPARK_CHAT' | 'ERNIE_CHAT' | 'YUANBAO_CHAT' | 'CHERRY_STUDIO_CHAT' | 'JAIS_CHAT' | 'STEPFUN_CHAT' | 'WRTN_CHAT' | 'MUSE_AI_CHAT': The AI Product the prompt was submitted to.
      - `prompt_use_case` string: What the user was trying to accomplish with the prompt.
      - `raw_prompt` string: The prompt text as the user submitted it. Absent when prompt capture is disabled.
      - `risk_analysis` PromptRiskAnalysis: Why the prompt was flagged: a summary and the specific exposures found.
        - `exposures` PromptExposure[]: Sensitive-data exposures found in the prompt text and attached files. Empty when the prompt was flagged for a reason other than sensitive data.
          - `exposure` 'FULL_NAME' | 'PHONE_NUMBER' | 'MEDICAL_RECORD_NUMBER' | 'PHYSICAL_ADDRESS' | 'SOCIAL_SECURITY_NUMBER' | 'CANADA_SIN' | 'ITIN' | 'EMPLOYER_ID' | 'DOB' | 'PRECISE_LOCATION' | 'DRIVERS_LICENSE' | 'US_PASSPORT_NUMBER' | 'EMAIL_ADDRESS' | 'FICO_CREDIT_SCORE' | 'ROUTING_NUMBER' | 'SWIFT_BIC_CODE' | 'IBAN' | 'CREDIT_CARD_PAN' | 'GENERIC_SECRET' | 'AWS_SECRET_KEY' | 'AWS_ACCESS_KEY_IDENTIFIER' | 'AWS_IAM_IDENTIFIER' | 'TWILIO_AUTH_TOKEN' | 'BEARER_TOKEN' | 'OPENAI_API_KEY' | 'GOOGLE_API_KEY' | 'GITHUB_CLASSIC_TOKEN' | 'GITLAB_TOKEN' | 'GITLAB_CELLS_TOKEN' | 'SLACK_TOKEN' | 'STRIPE_SECRET_KEY' | 'STRIPE_PAYMENT_TOKEN' | 'SENDGRID_API_KEY' | 'NPM_TOKEN' | 'PYPI_TOKEN' | 'DATADOG_PRIVATE_KEY' | 'SENTRY_DNS' | 'AZURE_CONNECTION_STRING' | 'DATABASE_CONNECTION_STRING' | 'PG_DATABASE_CONNECTION' | 'MONGODB_URI' | 'REDIS_CONNECTION_URI' | 'BASIC_AUTH_CREDENTIALS' | 'JWT_SESSION_TOKEN' | 'SAML_ASSERTION' | 'PRIVATE_KEY_PKCS8' | 'SSH_PRIVATE_KEY' | 'PGP_PRIVATE_KEY' | 'RSA_PRIVATE_KEY' | 'SOURCE_CODE' | 'OTHER': The kind of sensitive data that was exposed.
          - `file_name` string: Name of the attached file the exposure was found in. Only present when `source` is `FILE`.
          - `source` 'PROMPT' | 'FILE': Where the exposure was found: the prompt text or an attached file.
        - `prompt_risk_summary` string: Summary of the risk the prompt and its attachments pose.
      - `source` 'BROWSER_EXTENSION' | 'DEVICE_AGENT' | 'INTEGRATION': How Adaptive captured the prompt: with the browser extension, with the device agent, or through an integration with the AI product's vendor. Absent only for prompts captured before Adaptive recorded this.
      - `surface` 'BROWSER' | 'APP' | 'CLI' | 'UNKNOWN': Where the prompt was captured: in the browser, in a desktop app, or in a command-line tool. UNKNOWN when the prompt's source does not report where it was captured.
      - `user` User: The user the event concerns.
        - `email` string: Email address of the user.
        - `id` string: Unique Adaptive identifier for the user. Use it to fetch full user details from the Public API (GET /v2/users/{userId}).
  - `datacontenttype` string: media type of `data`
  - `id` string: unique id for event
  - `source` string: the event source; always `/adaptive`
  - `specversion` string: CloudEvents spec version
  - `subject` string: reserved for the event's primary entity id; currently never set, so always omitted
  - `time` string, date-time: when the event occurred
  - `type` 'ai_governance.prompt.flagged': the event type

## Acknowledgement `2XX`

Return any 2xx status to acknowledge the delivery. Timeouts, 408, 429, and 5xx responses other than 501 are retried; any other status is a permanent failure.

---

[API](https://skmtc.dev/adaptivesecurity/apis/adaptive-api.md) - [All operations](https://skmtc.dev/adaptivesecurity/apis/adaptive-api/llms.txt) - [OpenAPI document](https://skmtc.dev/adaptivesecurity/apis/adaptive-api/revisions/c367e8f673f3?raw)