Deepening our partnership with the UK AI Security Institute
Google DeepMind and UK AI Security Institute (AISI) strengthen collaboration on critical AI safety and security research
Continuously Augmented Discrete Diffusion model for Categorical Generative Modeling
Standard discrete diffusion models treat all unobserved states identically by mapping them to an absorbing [MASK] token. This creates an ‘information void’ where semantic information that could be inferred from unmasked tokens is lost between denoising steps. We introduce Continuously Augmented Discrete Diffusion (CADD), a framework that augments the discrete state space with a paired …
Read more “Continuously Augmented Discrete Diffusion model for Categorical Generative Modeling”
Implement automated smoke testing using Amazon Nova Act headless mode
Automated smoke testing using Amazon Nova Act headless mode helps development teams validate core functionality in continuous integration and continuous delivery (CI/CD) pipelines. Development teams often deploy code several times daily, so fast testing helps maintain application quality. Traditional end-to-end testing can take hours to complete, creating delays in your CI/CD pipeline. Smoke testing is …
Read more “Implement automated smoke testing using Amazon Nova Act headless mode”
Announcing MCP support in Apigee: Turn existing APIs into secure and governed agentic tools
Today, we expanded Google’s support for Model Context Protocol (MCP) with the release of fully-managed, remote MCP servers, giving developers worldwide consistent and enterprise-ready access to Google and Google Cloud services. This includes support for MCP in Apigee, which makes it possible for agents to use your secure, governed APIs and custom workflows cataloged in …
Semantic Mastery: Enhancing LLMs with Advanced Natural Language Understanding
Large language models (LLMs) have greatly improved their capability in performing NLP tasks. However, deeper semantic understanding, contextual coherence, and more subtle reasoning are still difficult to obtain. The paper discusses state-of-the-art methodologies that advance LLMs with more advanced NLU techniques, such as semantic parsing, knowledge integration, and contextual reinforcement learning. We analyze the use …
Read more “Semantic Mastery: Enhancing LLMs with Advanced Natural Language Understanding”
Real-world reasoning: How Amazon Nova Lite 2.0 handles complex customer support scenarios
Artificial intelligence (AI) reasoning capabilities determine whether models can handle complex, real-world tasks beyond simple pattern matching. With strong reasoning, models can identify problems from ambiguous descriptions, apply policies under competing constraints, adapt tone to sensitive situations, and provide complete solutions that address root causes. Without robust reasoning, AI systems fail when faced with nuanced …
From adoption to impact: Putting the DORA AI Capabilities Model to work
The 2025 State of AI-assisted Software Development report revealed a critical truth: AI is an amplifier. It magnifies the strengths of high-performing organizations and the dysfunctions of struggling ones. While AI adoption is now near-universal, with 90% of developers using it in their daily workflows, success is not guaranteed. Our cluster analysis of nearly 5,000 …
Read more “From adoption to impact: Putting the DORA AI Capabilities Model to work”
How AWS delivers generative AI to the public sector in weeks, not years
When critical services depend on quick action, from the safety of vulnerable children to environmental protection, you need working AI solutions in weeks, not years. Amazon recently announced an investment of up to $50 billion in expanded AI and supercomputing infrastructure for US government agencies, demonstrating both the urgency and commitment from Amazon Web Services …
Read more “How AWS delivers generative AI to the public sector in weeks, not years”
SO-Bench: A Structural Output Evaluation of Multimodal LLMs
Multimodal large language models (MLLMs) are increasingly deployed in real-world, agentic settings where outputs must not only be correct, but also conform to predefined data schemas. Despite recent progress in structured generation in textual domain, there is still no benchmark that systematically evaluates schema-grounded information extraction and reasoning over visual inputs. In this work, we …
Read more “SO-Bench: A Structural Output Evaluation of Multimodal LLMs”
Using MCP with Web3: How to secure agents making blockchain transactions
At Google Cloud, we sit at a unique intersection of two transformative technologies: AI and Web3. The rise of AI agents capable of interacting with blockchains opens up a world of automated financial strategies, fast payments, and more complex scenarios like executing complex DeFi operations and bridging assets across multiple chains. However, the practical viability …
Read more “Using MCP with Web3: How to secure agents making blockchain transactions”