AI content labels

Teams offers three kinds of labels that your agent can apply to its messages to help users understand their contexts:

A standardized AI Generated label clearly marks the message as a product of AI.
Citations enable agents to direct users to information sources used to create their messages.
Sensitivity labels enable users to understand the confidentiality of the agent message.

Note

AI content labels are available for agents in personal chats, group chats, and channels. They are available in Government Community Cloud (GCC), GCC High, and Department of Defense (DoD) environments.

AI Generated label

Call AddAIGenerated() on a message activity before sending it to display a platform-standard AI Generated label next to the agent's name in the message header. Hovering over the label displays the tooltip AI-generated content may be incorrect..

async Task SendAILabel(IContext context)
{
    await context.Send(new MessageActivity
    {
        Text = "Hi, this message is generated by AI"
    }.AddAIGenerated());
}

Call add_ai_generated() on a message activity before sending it to display a platform-standard AI Generated label next to the agent's name in the message header. Hovering over the label displays the tooltip AI-generated content may be incorrect..

@app.on_message
async def add_ai_label(ctx: ActivityContext[MessageActivity]):
    await ctx.send(
        MessageActivityInput(
            text="Hi, this message is generated by AI",
        ).add_ai_generated()
    )

Call addAiGenerated() on a message activity before sending it to display a platform-standard AI Generated label next to the agent's name in the message header. Hovering over the label displays the tooltip AI-generated content may be incorrect..

app.message(async ({ send }) => {
  await send(new MessageActivity("Hi, this message is generated by AI").addAiGenerated());
});

Citations

Citing sources in agent messages helps users ask follow-up questions or conduct independent research. Cite data sources like files, messages, emails, and work items to provide valuable insights. Citations are crucial for agents using techniques like Retrieval-Augmented Generation (RAG).

Citations

Modal window

Citations in your agent's messages can include the following:

In-text citations denote the citation numbers added to the agent message in the [#] format, each corresponding to a reference. A citation can be inserted anywhere within the text.
Details of the citation reference include the title, icon, keywords, abstract, hyperlink, sensitivity information, and a button to open a modal window with additional content. References appear as pop-up windows for each in-text citation.
Sensitivity labels to citations indicate the confidentiality of the citation content referenced and aren't added automatically. To add sensitivity labels for citations, see add sensitivity label.
Modal window with additional content renders an Adaptive Card without any interactive items.

Note

A maximum of 20 citations are displayed in a message.
Citations with Adaptive Cards are available in public developer preview.
Adaptive Cards aren't rendered in the citation pop-up window. However, Adaptive Cards can be rendered in the agent's message or in the citation's modal window accessible from the pop-up window.

Add citations

If you're using Teams SDK to build your agent, Use addCitation() to include in-text references and citation metadata in your message. Following is an example code snippet:

app.message(/citation/i, async ({ send }) => {
  const card = new AdaptiveCard(new TextBlock("Adaptive Card text"))
    .withOptions({ version: "1.6" });
  const appearance: CitationAppearance = {
    name: "AI messages agent",
    url: "https://example.com/claim-1",
    abstract: "Excerpt description",
    text: JSON.stringify(card),
    keywords: ["keyword 1", "keyword 2", "keyword 3"],
    icon: "Microsoft Word",
  };
  await send(
    new MessageActivity("Hey I'm a friendly AI agent. This message is generated through AI [1]")
      .addCitation(1, appearance)
  );
});


async Task SendCitations(IContext context)
{
var message = new MessageActivity
{
Text = "Hey I'm a friendly AI agent. This message is generated through AI [1]"
};

message.AddCitation(1, new CitationAppearance
{
    Name = "AI messages agent",
    Url = "https://example.com/claim-1",
    Abstract = "Excerpt description",
    Keywords = new List { "keyword 1", "keyword 2", "keyword 3" },
    Icon = CitationIcon.MicrosoftWord
});

await context.Send(message);
}

@app.on_message_pattern(re.compile(r"citation", re.IGNORECASE))
async def add_citations(ctx: ActivityContext[MessageActivity]):
    await ctx.send(
        MessageActivityInput(
            text="Hey I'm a friendly AI agent. This message is generated through AI [1]",
        ).add_citation(
            position=1,
            appearance=CitationAppearance(
                name="AI messages agent",
                url="https://example.com/claim-1",
                abstract="Excerpt description",
                text=AdaptiveCard(version="1.6", body=[TextBlock(text="Adaptive Card text")]).model_dump_json(by_alias=True, exclude_none=True),
                keywords=["keyword 1", "keyword 2", "keyword 3"],
                icon=CitationIconName.MICROSOFT_WORD,
            ),
        )
    )

Property	Type	Required	Description
`citation`	Object	✔️	Details of the citation.
`citation.@type`	String	✔️	Object of the citation. Allowed value: `Claim`
`citation.position`	Integer	✔️	Displays the citation number. This value must be unique for every citation.
`citation.appearance`	Object	✔️	Information about the appearance of the citation.
`citation.appearance.@type`	String	✔️	Object of the citation appearance. Allowed value: `DigitalDocument`
`citation.appearance.name`	String	✔️	Title of the referenced content. Maximum characters: 80
`citation.appearance.url`	String		URL of the referenced content.
`citation.appearance.abstract`	String		An abstract of the referenced content. Maximum characters: 160
`citation.appearance.text`	String		A stringified Adaptive Card with additional information about the citation. It renders within the modal window accessible from the pop-up window.
`citation.appearance.keywords`	Array		Keywords from the referenced content. You can't add more than three keywords. Each keyword can only contain 28 characters.
`citation.appearance.encodingFormat`	String		The encoding format of the `citation.appearance.text` field. Allowed value: `application/vnd.microsoft.card.adaptive`
`citation.appearance.image`	Object		Information about the citation's icon.
`citation.appearance.image.@type`	String	✔️	The object of the citation icon. Must be `ImageObject`.
`citation.appearance.image.name`	String	✔️	The name of the predefined icon. It renders the citation icon in the details of the citation reference. Allowed values: `Microsoft Word`, `Microsoft Excel`, `Microsoft PowerPoint`, `Microsoft OneNote`, `Microsoft SharePoint`, `Microsoft Visio`, `Microsoft Loop`, `Microsoft Whiteboard`, `Source Code`, `Sketch`, `Adobe Illustrator`, `Adobe Photoshop`, `Adobe InDesign`, `Adobe Flash`, `Image`, `GIF`, `Video`, `Sound`, `ZIP`, `Text`, `PDF`

After you enable citations, the agent message includes in-text citations and references. The in-text citations display the reference details when users hover over the citation.

Error handling

Error code	Description
400	Multiple root message entities found under `entities` array.
400	Error parsing message entity from `entities` array.
400	Agent message with more than 20 citations.
400	The `appearance` object is empty.
400	Error while parsing citation entity with ID: X.

Sensitivity label

Agent responses might contain confidential information or be accessible only to certain individuals within the organization. Add a sensitivity label to help users identify the confidentiality of a message, enabling them to exercise caution when sharing it.

Screenshot shows an AI-powered agent with sensitivity label in the Teams desktop client.

Note

Add a sensitivity label to your agent's messages only when they contain sensitive information.

Add sensitivity label

For agents built using Teams SDK, add a sensitivity label to your agent message by modifying the message to include usageInfo in the entities object.

The following code snippet shows how to add sensitivity labels to both agent messages and citation reference:

await context.sendActivity({
  type: ActivityTypes.Message,
  text: `Hey, I'm a friendly AI agent. This message is generated through AI [1]`,
  entities: [
    {
      type: "https://schema.org/Message",
      "@type": "Message",
      "@context": "https://schema.org",
      usageInfo: {
        "@type": "CreativeWork",
        name: "Sensitivity title",
        description: "Sensitivity description",
      },
    },
  ],
});

await context.sendActivity({
  type: ActivityTypes.Message,
  text: `Hey, I'm a friendly AI agent. This message is generated through AI [1]`,
  entities: [
    {
      type: "https://schema.org/Message",
      "@type": "Message",
      "@context": "https://schema.org",
      usageInfo: {
        "@type": "CreativeWork",
        name: "Sensitivity title",
        description: "Sensitivity description",
      },
      citation: [
        {
          "@type": "Claim",
          position: 1,
          appearance: {
            "@type": "DigitalDocument",
            name: "AI agent",
            url: "https://example.com/claim-1",
            usageInfo: {
              "@type": "CreativeWork",
              "@id": "sensitivity1",
              name: "Sensitivity title",
              description: "Sensitivity description",
            },
          },
        },
      ],
    },
  ],
});

Property	Type	Required	Description
`usageInfo.@type`	String	✔️	Enables the sensitivity label in the agent message.
`citation.usageInfo.@id`	String	✔️	Enables the sensitivity label in the citation reference. It's required when adding sensitivity label to citation reference.
`usageInfo.name`	String	✔️	Specifies the title of the sensitivity label.
`usageInfo.description`	String		Specifies the pop-up window message that appears when a user hovers over the sensitivity label.

After you add the sensitivity label, your agent message displays a shield icon. Users can hover over the icon to see a disclaimer about the message's sensitivity.

Error handling

Error code	Description
400	Multiple root message entities found under `entities` array.
400	Error parsing message entity from `entities` array.
400	Citation level `usageInfo.@id` value doesn't match the message level `usageInfo.@id` in at least one instance.
400	There are multiple citation-level `usageInfo` properties with the same `@id`, but their `name` and `description` properties are different.

Code sample

Sample Name	Description	Node.js	.NET	Python
Teams conversation agent	This sample app displays AI labels, citations, and sensitivity labels in messages.	View	View	View

Feedback

Was this page helpful?

Last updated on 2026-09-28

AI content labels

AI Generated label

Citations

Add citations

Error handling

Sensitivity label

Add sensitivity label

Error handling

Code sample

See also

Feedback

Additional resources