Google briefly rolled out Nano Banana 2 inside the web version of Google Earth, giving users a new "Create image" button that could rewrite real places with AI-generated visuals tied to satellite imagery, 3D terrain, and camera parameters. The feature was pitched for use cases such as reconstructing historical sites, producing landmark infographics, previewing architectural concepts, and imagining future cityscapes without leaving Google Earth. The rollout did not last. According to the source article, the tool was pulled less than a day after launch after highly realistic results triggered strong objections from experts. Google then removed the feature and is expected to relaunch it after adding stronger safeguards. The report also outlined how the system works. It said Nano Banana 2 relies on Gemini’s multimodal and search-grounding capabilities, uses live geospatial constraints rather than freeform text-to-image generation, and can render up to 2K or 4K output with sharper text. At the same time, testers and media outlets questioned how reliable the tool would be for professional use, especially since it does not support direct use in Street View and may not preserve strict spatial geometry for developers or architects.

Google has introduced Nano Banana 2 Lite, a new enterprise-focused image model that generates 1K images in 4 seconds and costs $0.034 per 1,000 images through its API stack.

Anthropic mistakenly removed Claude Code from its Pro plan during an A/B test, sparking massive cancellations and pushing developers to OpenAI's Codex. The rival tool now boasts 4 million weekly actives, up 1 million in two weeks.

Google DeepMind releases Nano Banana 2, merging Pro-level quality with Flash speed. Supports 4K resolution, maintains up to 5 characters or 14 objects consistently. Integrated across Gemini, Google Search, and more.

Decrypt published a detailed walkthrough showing how sellers can build short product ads for TikTok and YouTube with AI tools that are free or close to free. The workflow starts with a clean product image, moves through ChatGPT for product-consistent model shots and a 10-second script in JSON, then uses Google Flow with Gemini Omni to generate talking video clips, and ends with CapCut for trimming and subtitles. The article also breaks down pricing: Google Flow gives non-subscribers 50 free daily credits for Veo 3.1, while Omni requires a paid Google AI plan, though Google also offers Omni at no cost inside YouTube Shorts and YouTube Create for users 18 and older. Decrypt included platform rules as well. TikTok and YouTube both allow AI-generated promotional videos, but they require labels for realistic AI media and separate disclosure for commercial relationships. The piece closes with a reminder that cheap tools do not mean easy sales, citing Third Eye Insights president Camille Moore, who said that more than half of the 803,500 TikTok Shop stores operating in the U.S. last year recorded zero sales.

Google has released two generative AI models: Gemini Omni Flash for video generation and Nano Banana 2 Lite for image generation. The former supports multimodal input and conversational video editing; the latter achieves 4-second image generation with low-cost batch production. They can be chained to create an end-to-end AIGC workflow targeting e-commerce, decoration, and short video verticals.