Our Summary:Google has made its SynthID detector available to anyone through a new website, allowing users to check for invisible watermarks in images, videos, and audio. The detector recognizes watermarks from partner companies, including OpenAI, ElevenLabs, and Nvidia. This move aims to provide a more transparent and verifiable way to identify AI-generated content.

Our Summary:The Chan Zuckerberg Biohub has teamed up with Google DeepMind and the US government to develop an AI-powered virtual cell, aiming to predict how real cells behave in response to various stimuli. The $1.8 billion initiative, which includes $500M from the US Department of Energy and $300M from Meta, Google DeepMind, and Isomorphic Labs, will produce large multimodal biological datasets and open them to researchers worldwide.

Our Summary:Google has launched Playground, a browser-based platform that lets US adults build and test custom games without writing a line of code. The platform uses AI prompts to generate games, reducing the obstacles to creating games and providing a conversational interface for users to direct the game development process. This move is part of Google's efforts to provide access to game development without technical training, following its previous AI-driven projects such as YouTube Playables Builder and Project Genie.

Our Summary:Black Hills Corp. has signed a power service agreement with Google to support a new data center in Cheyenne, Wyoming, with a total investment of $1.8 billion between 2027 and 2029. The utility will generate 564 MW of new natural gas capacity to support Google's grid service needs, with a total resource mix of 2.7 GW.

Our Summary:Google has launched its new image model, Nano Banana 2.1, which offers better performance than its predecessor at a significantly lower cost. The new model is now available for use, and its price has been reduced by half compared to its predecessor.

Our Summary:Constellation Energy has entered into a partnership with Google to upgrade 11 nuclear reactors in Illinois, Pennsylvania, and New Jersey, valued at over $4.3 billion. The deal includes long-term power contracts and a technology tie-up with Alphabet, and is expected to add 890 MW of new nuclear generating capacity to the PJM Interconnection grid.

Our Summary:Google has released EmbeddingGemma 2, a 740 million parameter open-weight model that enables multimodal embeddings. The model ships under an Apache 2.0 license, allowing commercial use. It can handle text, images, video, and audio, and has a 4x larger context window than the original EmbeddingGemma.

Our Summary:Google has unveiled EmbeddingGemma 2, a lightweight multimodal model that outperforms larger models and handles multiple tasks. This successor to the original EmbeddingGemma 2 is designed for on-device efficiency, making it suitable for privacy-sensitive applications and offline use cases.

Our Summary:A post on X claimed Google's Nano Banana 2.1 is now available, offering high-quality image generation at lower prices. However, Google has not confirmed any rollout of the model, and it is missing from official documentation. The listing was briefly spotted in Google Flow, but it was pulled shortly after users reported it.

Our Summary:Mysten Labs and Google Cloud have unveiled the Verifiable Agent Arbiter (VAA), a solution to prove AI agents' actions and prevent unauthorized behavior. The VAA splits agent activity into two tracks, storing telemetry in Google Cloud Storage and cryptographic proofs on the Sui blockchain.

Our Summary:Google is reducing the features available to free Gemini users, starting October 9. Free users will only have access to the Gemini 3.5 Flash-Lite model, while paid users will lose access to Gemini Pro. The changes aim to encourage users to upgrade to paid plans.