The AI tool, known as Imagen 3, will be integrated into the Gemini platform and will initially be accessible to Gemini Advanced, Business, and Enterprise users in English.
The AI tool, known as Imagen 3, will be integrated into the Gemini platform and will initially be accessible to Gemini Advanced, Business, and Enterprise users in English.
Dave Citron, a senior director of product at Gemini, stated that Google has enhanced the technology and refined its evaluation methods, including conducting red-teaming exercises to test for vulnerabilities.
Earlier this year, Google halted the feature after it inaccurately rendered historical images, sparking widespread criticism on social media.
For example, a request for an image of a German soldier from 1943 resulted in a racially diverse group in German uniforms, and a query for a medieval British king produced images including a female ruler.
The updated Imagen 3 will exclude the creation of photorealistic identifiable individuals, depictions of minors, or images that are overly violent, gory, or sexual.
Citron emphasized that while not every generated image might be perfect, Google is committed to continuously improving the tool based on user feedback. The rollout will be gradual, aiming to expand availability to more users and languages.
TechEchelon Staff bylines are produced collectively by the newsroom for short, breaking, and wire-style coverage. Longer-form reporting is published under the responsible reporter's name.
More from the Staff →The Justice Department's Civil Rights Division announced that OpenAI and former subsidiary Statsig signed a $3.2 million settlement covering alleged immigration law violations, placing both companies under three years of federal oversight of their green-card hiring practices.
Google announced a sweeping AI leadership shake-up on Wednesday, with chief scientist Jeff Dean departing after 27 years to start a new company and Demis Hassabis stepping back from the CEO role at Google DeepMind to become its chairman and Alphabet's new chief scientist.
The Trump administration's AI model-testing framework excludes open-source models entirely, sets a 30-day government review window for closed models, and leaves key terms like "national security risk" undefined, raising questions about the initiative's practical reach.
Written for readers who already know the basics — markets, AI, and the policy decisions that shape both.