This article is for Google Workspace administrators. Gemini users: Learn how Google protects you when you use Gemini Apps or the Gemini side panel with Gmail or Docs editors.
Gemini has become a powerful collaborative partner for Google Workspace business users. Thousands of organizations rely on Gemini to create content, summarize data, brainstorm, and increase productivity. At the same time, emerging threats have begun to target generative AI platforms.
To defend against these threats, Google uses a robust framework of security measures. This article explains what happens when those security measures are applied, and how they might affect the user experience with Gemini for Google Workspace and Gemini Apps.
On this page
- About emerging gen AI threats
- How Gemini helps protect users
- What users can expect
- What users should do
About emerging gen AI threats
Emerging gen AI threats include suspicious URLs and malicious prompts. Similar to malware and phishing scams, these threats can be difficult for users to spot. Attackers try to hide malicious prompts in legitimate-looking email messages or documents, with the goal of getting gen AI to do something unintended or harmful.
For example, malicious prompts might ask gen AI to share personal information, or perform unauthorized tasks.
How Gemini helps protect users
When users ask Gemini to summarize content such as email messages or documents, Gemini analyzes that content for threats before taking action. When Gemini identifies threats in content, it excludes that content from being used to generate a response.
For example:
- If you ask Gemini to summarize email messages, and one of those messages contains malicious prompts, that message is excluded from Gemini’s summary.
- If you prompt Gemini to use content in files, such as docs, Gemini checks those files for suspicious links or prompts before processing your request.
What users can expect
When threats are detected, Gemini displays the following types of notifications to users:
| Threat | Notification to users |
|---|---|
| All email messages or documents referenced in a prompt are suspicious, or a conversation contains malicious instructions attempting a prompt injection attack | A security risk was identified and blocked |
| Some messages or documents are suspicious | Some content was excluded for security reasons |
What users should do
Users should send feedback to Google if they’re concerned about gen AI security risks, or if they think Gemini made a mistake.
For instructions, go to:
Related topics
- Manage Gemini for Google Workspace for your organization
- Gemini for Google Workspace FAQ
- Tips for using Gemini in your job or role
- Learn how Google protects gen AI users in:
Google, Google Workspace, and related marks and logos are trademarks of Google LLC. All other company and product names are trademarks of the companies with which they are associated.
