Google has announced that its Gemini AI model, powered by the Imagen 3 engine, will soon enable Google Workspace users to generate images featuring people.
This reintroduction comes after the feature was suspended in February 2024 due to concerns over inaccuracies and biases in visual outputs.
To mitigate potential misuse, Google has introduced safeguards aimed at ensuring responsible use.
Gradual Rollout with Enhanced Security Measures
The new feature rollout follows extensive testing and early access that began in August 2024. During this period, Google focused on reducing risks related to deepfakes and inaccurate representations.
One key addition is SynthID, a tool designed to embed a non-visible watermark in AI-generated images, ensuring traceability and authenticity.
Users will be able to access this feature through Gemini’s mobile apps, Gemini Advanced, and side panels in Google Docs, Sheets, Drive, Slides, and Gmail.
Access is limited to paid subscribers, with those on free tiers excluded from generating images featuring people. The global rollout is expected to conclude by March 1, 2025.
| Feature | Previous Image Generation (Suspended) | Gemini with Imagen 3 |
|---|---|---|
| Status | Suspended (Feb 2024) | Reintroduced & Active (Rollout until Mar 1, 2025) |
| Accuracy & Bias | Concerns over inaccuracies and biases | Enhanced prompt-following & improved accuracy |
| Security Safeguards | Lacked robust safeguards | Introduces SynthID watermark for traceability |
| Access | Unavailable due to suspension | Accessible via Gemini mobile apps, Advanced, and Workspace side panels (paid subscribers only) |
| Creative Capabilities | Limited creative control | Supports hyper-realistic visuals, abstract art, animations, customizable outputs (aspect ratios, image count) |
| Developer Access & Pricing | Not available | Via Gemini API at $0.03 per image with configurable prompt settings |
| Additional Features | Basic image generation | Enhanced personalization via “Gems” for custom AI experts and user control improvements |
AI-Powered Enhancements for Google Workspace
In addition to image generation, Google is launching several AI-driven updates to enhance the Workspace experience.
One key update allows users to insert Gemini-generated chatbot responses directly into Gmail drafts, streamlining communication.
Enterprise users of Google Chat will also benefit from new quick commands, enabling developers to trigger app functions without manually typing slash commands.
These commands can be configured through the Chat API page and accessed via a plus button near the compose box.
Imagen 3 – Advancing Creative Image Generation
Imagen 3 offers enhanced capabilities for generating visually compelling, artifact-free images. It supports a wide range of creative content, from hyper-realistic visuals to abstract art and animations.
With improved prompt-following, users can efficiently translate ideas into high-quality visuals.
The model’s flexibility allows users to control specifications such as aspect ratios and the number of images, catering to diverse creative needs.
Imagen 3 is positioned as a valuable tool for both professional and personal creative applications.
Developer Access and Pricing
Developers can access Imagen 3 through the Gemini API, with the service initially available to paid users at $0.03 per generated image. Google has indicated plans to extend this access to free-tier users in the future.
Developers also have options to configure prompt settings and the number of image generations, enabling tailored creative solutions.
To facilitate seamless integration, Google has provided sample code snippets to help developers adopt Imagen 3 capabilities efficiently.
Ethical Safeguards and Responsible Image Generation
Google has integrated SynthID to watermark AI-generated images, helping to distinguish them from real photographs and reducing potential misuse, such as deepfake creation.
To further ensure responsible use, Imagen 3 adheres to stringent design principles that prevent the generation of visuals depicting identifiable individuals, minors, or inappropriate content. Feedback from early users will guide future improvements.
Introducing Personalized AI Experts – Gems
A new feature called “Gems” allows users to create custom AI experts on various topics. Available to Gemini Advanced, Business, and Enterprise users, these AI assistants provide tailored support for tasks like brainstorming, writing, and coding.
Pre-made Gems being rolled out include –
- Learning Coach – Simplifying complex topics.
- Brainstormer – Generating creative ideas for events and projects.
- Career Guide – Offering skill development and career growth advice.
- Writing Editor – Providing constructive feedback for content improvement.
- Coding Partner – Assisting with project development and coding support.
Enhanced Image Generation Experience
Google emphasizes user control throughout the creative process. Users dissatisfied with generated images can refine outcomes by providing additional instructions.
The phased rollout of image generation for people will focus on technical refinement and user evaluation, initially available to select business and enterprise users.
Frequently Asked Questions
How does Gemini with Imagen 3 differ from previous image generation features?
Unlike earlier versions—which were suspended due to inaccuracies and biases—Gemini with Imagen 3 incorporates enhanced prompt-following and improved image quality, along with new safeguards like SynthID to ensure traceability and responsible use.
What privacy safeguards are in place when generating images of people?
Google has implemented SynthID to embed a non-visible watermark in images, ensuring traceability and authenticity. Strict design principles also prevent the generation of visuals that compromise privacy or depict identifiable individuals inappropriately.
Will the feature support regional and language-specific customization?
While the current rollout focuses on core functionality, future updates may include regional and language-specific customization to cater to diverse global users.
How can developers integrate the Gemini API for Imagen 3?
Developers can access Imagen 3 via the Gemini API, which provides sample code snippets and configurable options for prompt settings and image generation volume. This integration is designed to be straightforward, allowing seamless adoption into existing applications.
Is there an option for batch image processing for enterprise users?
Although the current pricing is set at $0.03 per image for individual requests, Google has indicated that future updates may include batch processing capabilities, especially for enterprise users requiring high-volume, efficient image generation.
