Google has reactivated its AI image generation feature with enhanced safeguards after a previous controversy over inaccurate and historically problematic images. The updated system, now powered by the latest Imagen 3 model, is set to make a comeback on the Gemini AI platform, starting August 28.
New Guidelines and Features: The refreshed Imagen 3 model introduces stricter controls to prevent issues encountered earlier this year.
The new features include:
No Photorealism: The model will avoid creating highly realistic images of people.
Content Restrictions: It will not generate images of identifiable individuals, minors, or any graphic, violent, or sexual content.
Language Limitation: Currently, the image generation feature will only be available in English and for users of Gemini Advanced, Business, and Enterprise tiers.
Background: In February, Google had to suspend its image generation feature after it produced images that were both diverse and historically inaccurate, such as portraying historical figures inappropriately. This led to a backlash accusing the company of overemphasizing diversity at the expense of historical accuracy. High-profile critics, including Elon Musk, amplified the controversy by criticizing the approach.
Updates and Future Plans: Google’s latest improvements aim to address past issues while enhancing the user experience. The company also unveiled “Gems,” adaptable chatbots designed for diverse tasks like code review, language tutoring, or writing support. This feature is part of Google’s broader strategy to refine its AI tools and better meet user needs.
