AI News — Wednesday, July 29, 2026
Cyera has agreed to acquire Oasis Security for $1 billion, signaling a major investment in securing the rapidly growing ecosystem of AI agents.
A new 'frontier intelligence' model, Kimi K3, has been released, garnering significant attention for its capabilities.
Google has unveiled the first redesign of its search box in 25 years, a move widely interpreted as a foundational change to better integrate AI-powered search experiences.
Spur, a startup specializing in bot detection, has raised $200 million from Insight, highlighting increasing concerns and investments in combating automated threats.
Google has announced significant updates to its Gemini API Managed Agents, introducing Gemini 3.6 Flash, new hooks, and other enhancements for developers.
A new cybersecurity threat called 'Slopsquatting' has emerged, exploiting AI hallucinations to launch supply chain attacks.
OpenAI has published insights into how agentic AI is transforming scientific computing, suggesting a future where AI agents play a pivotal role in research and discovery.
Developers are being urged to immediately audit any AI agents granted write access to public repositories due to potential security vulnerabilities.
Google highlights five practical ways its AI Mode in Search can help users engage more effectively with the real world, from planning to problem-solving.
New research explores how relevance can be redefined to guide AI agents in interacting with information corpora, enhancing agentic search capabilities.
Researchers are re-evaluating classifier-free guidance techniques within on-policy diffusion distillation, aiming to improve the efficiency and quality of diffusion models.
New research introduces HiFi-UMI, a method for learning effective robotic manipulation policies solely from high-fidelity unstructured multi-modal interaction (UMI) data.
A new paper delves into the underlying 'physics' of multi-turn, long-horizon planning for AI agents, exploring pre-training and post-training via distillation.
ClinFusion introduces a novel vision-centric multimodal large language model system designed for comprehensive understanding in the medical domain.
Mage-VL is presented as an efficient, codec-native streaming multimodal foundation model, promising advancements in real-time multimodal processing.