Google is addressing the growing threat of indirect prompt injection attacks on generative AI systems, which involve hidden malicious instructions in external data sources. Their layered security strategy for the Gemini platform includes advanced content classifiers, security thought reinforcement, markdown sanitization, user confirmation mechanisms, and end-user security notifications to enhance protection against such attacks.

Figma is enhancing its platform by integrating Google's generative AI models, including Gemini 2.5 Flash and Imagen 4, to significantly speed up image generation and editing, achieving a 50% reduction in latency. The integration maintains enterprise governance with opt-out controls for AI content training and ensures image quality with built-in watermarking for provenance. This move aims to streamline design processes while adhering to organizational standards and expectations.

Google’s Veo 3.1 is generating excitement in the generative video AI landscape, particularly as it aims to compete with OpenAI's Sora 2. With improved visual fidelity and audio tracks, early tests suggest Veo 3.1 produces more nuanced and detailed outputs compared to its predecessor, Veo 3. The model is set for a broader release soon, promising to enhance video production for creators and marketers alike.

Google Labs has collaborated with Brooklyn design duo Wade and Leta to create "Reflection Point," an interactive mirrored maze sculpture unveiled at Rockefeller Center. Utilizing Google's generative AI tool, Whisk, the artists enhanced their creative process and aimed to make the artwork accessible to all. The sculpture will be on display at Center Plaza until July 20.

Google DeepMind's Veo 3, unveiled at the I/O developer conference, significantly enhances generative AI video quality by integrating sound effects and realistic dialogue, making it challenging to distinguish from human-created content. As generative AI technology advances, concerns grow regarding the potential prevalence of AI-generated videos and the implications for human creators, emphasizing the need for media literacy to identify such content.

Google DeepMind has introduced a cloud-free on-device VLA model for robotics, enhancing the autonomy and reliability of physical robots. This innovation allows robots to quickly adapt to their environments without relying on cloud processing, marking a significant advancement in generative AI applications in robotics.

Adobe has integrated Google's Gemini 2.5 Flash Image into its Firefly and Adobe Express platforms, allowing creators enhanced capabilities in visual generation. To celebrate this launch, Adobe is offering unlimited generations for Firefly and Creative Cloud subscribers until September 1, while new users on the free plan can access up to 20 generations. Despite the advancements in image quality and editing control, Google acknowledges that the model has limitations, including potential inaccuracies in generated content.

Google is set to launch a public experiment for its Genie 3 model, allowing users to create and explore virtual worlds through simple prompts. This initiative aims to enhance user interaction with the technology, following feedback from the initial reveal, and will likely target early adopters and researchers interested in generative AI. No specific release date has been provided yet.

Promise, a generative AI-powered studio founded by industry veterans, has secured a strategic investment from Google’s AI Future Fund and other investors to enhance its AI-assisted productions. The startup is set to showcase its innovative projects, including the film "NinjaPunk," while promoting the use of generative AI to enhance creativity in Hollywood rather than replace it.