The day's AI news, briefed
The Daily Prompt
Our AI editor reads each day's AI and robotics stories, pulls out why they matter, and links every original source — so you can keep up in minutes.
Join over 250,000 subscribers getting the latest AI news
Latest AI & Robotics News Today
With the Rise of AI, Cisco Sounds an Urgent Alarm About the Risks of Aging Tech
Aging digital infrastructure increases cybersecurity risks, especially as generative AI facilitates exploitation of vulnerabilities. Addressing this issue is crucial for safeguarding network security worldwide.
- Old equipment like routers and network switches is often left unmonitored, creating security vulnerabilities.
- Cisco's "Resilient Infrastructure" initiative aims to raise awareness about the risks of outmoded technology.
- Research showed the UK and US face the highest risks from legacy technology in critical sectors.
- Cisco plans to implement new warnings for products nearing end of life to enhance digital security.
- Generative AI accelerates attackers' ability to exploit known vulnerabilities, raising the urgency for infrastructure upgrades.
Cisco, founded in 1984, is a leading network technology company whose equipment is integral to global digital infrastructure. The company has long called attention to the risks of aging technology, stressing the need for regular updates and investment in security. The new initiative emphasizes the importance of treating these risks as critical for organizational leadership.
Read the original ↗
Lightweight LLM powers Japanese enterprise AI deployments
The deployment of a lightweight LLM (Large Language Model) in Japanese enterprises signifies advancements in AI efficiency and accessibility, particularly in resource-constrained environments.
- A lightweight LLM is being utilized to enhance AI deployments in Japanese enterprises.
- This technology allows for operation on a single GPU, making it more accessible for various applications.
Large Language Models (LLMs) are advanced AI systems that understand and generate human-like text. Previous attempts to deploy such models in enterprise settings often required significant computational resources. The current development focuses on making AI more feasible for organizations with limited infrastructure.
Based on the headline only — the full article could not be fetched.
Read the original ↗
Kandinsky 5.0: A Family of Foundation Models for Image and Video Generation
The introduction of Kandinsky 5.0 models may significantly impact the capabilities and accessibility of generative AI applications for high-resolution image and video synthesis.
- Kandinsky 5.0 is a family of generative models designed for high-resolution image and short video generation.
- The framework includes three models: Image Lite, Video Lite, and Video Pro, with parameters ranging from 2B to 19B.
- The models utilize advanced training techniques and optimizations for improved performance and speed.
Kandinsky 5.0 represents a significant advancement in generative models for visual media, building on previous versions to achieve higher quality outputs. The models are publicly available for research purposes, aiming to enhance accessibility and development in the field of generative AI.
Read the original ↗
vLLM vs TensorRT-LLM vs HF TGI vs LMDeploy, A Deep Technical Comparison for Production LLM Inference
Understanding the differences among various LLM (Large Language Model) inference systems could influence decisions on efficiency and performance in AI deployments.
- This article compares vLLM, TensorRT-LLM, HF TGI, and LMDeploy in the context of production LLM inference.
- The focus is on deep technical aspects of each framework for AI applications.
Large Language Models (LLMs) are essential for many AI applications, requiring efficient inference frameworks for practical deployment. Previous comparisons have highlighted varying performance metrics among different LLM systems. Observers should watch for innovations that emerge from these competing technologies.
Based on the headline only — the full article could not be fetched.
Read the original ↗
Is Gemini 3 Better than GPT 5.1?
Understanding the differences between Gemini 3 and GPT 5.1 highlights their respective strengths, influencing users' choices for AI tools based on specific tasks and use cases.
- Gemini 3 integrates deeply with Google's ecosystem and emphasizes multimodality, making it advantageous for users entrenched in Google services.
- GPT 5.1 offers improved reasoning and conversational capabilities, prioritizing user-friendly interaction and flexible operational modes.
- Gemini 3 outperformed GPT 5.1 in coding tasks but struggled with complex image processing, while GPT 5.1 demonstrated better accuracy in image analysis.
Gemini 3 is Google's latest large language model aimed at enhancing various applications within its ecosystem. GPT 5.1 is OpenAI's recent update designed to improve task performance and user engagement. Both models cater to different user needs, making them valuable tools in AI deployment.
Read the original ↗
OmniParser for Pure Vision Based GUI Agent
The introduction of OmniParser could revolutionize user interface interactions for AI agents, leading to improved automation in various applications across different operating systems.
- OmniParser enhances the capabilities of GPT-4V by parsing user interface screenshots into actionable elements.
- It significantly improves GPT-4V's performance on benchmarks such as ScreenSpot, Mind2Web, and AITW.
- The method includes a curated dataset for detecting interactable icons and understanding interface semantics.
OmniParser is a new technique designed to improve how AI models interact with graphical user interfaces. The technology builds on recent advancements in vision language models and addresses limitations in current parsing capabilities. Future developments will focus on the potential release of the datasets and finetuned models used in this research.
Read the original ↗
Nvidia CEO Dismisses Concerns of an AI Bubble. Investors Remain Skeptical
Nvidia's ability to sustain demand amidst investor skepticism highlights the challenges of the AI industry's growth and the potential impact on market stability.
- Nvidia CEO Jensen Huang addressed concerns about an AI bubble during a recent earnings call, asserting high demand for their products.
- Nvidia reported record sales of $57 billion and a profit of nearly $32 billion, with expectations of $65 billion in the current quarter.
- The company has over $500 billion in unfilled orders and faces mixed investor reactions following a 10% decline in shares after hitting an all-time high.
- Huang emphasized that Nvidia's partnerships with AI companies like OpenAI and Anthropic are vital for technical support and growth.
Nvidia has become a leading provider of graphics processing units (GPUs), essential for running AI applications. The company has seen rapid growth due to increased demand since the launch of ChatGPT. Investors are now scrutinizing its sustainability and the broader implications for the AI market.
Read the original ↗
Trump Takes Aim at State AI Laws in Draft Executive Order
This draft executive order could reshape the landscape for AI regulation in the U.S. by limiting state control and shifting focus to a federal level, impacting how artificial intelligence is developed and managed.
- President Trump is preparing an executive order to challenge state AI regulations through lawsuits and federal funding restrictions.
- The draft order would establish an AI Litigation Task Force within the U.S. Justice Department to sue states over AI laws.
- It specifically targets recently enacted AI safety laws in California and Colorado, which require transparency from AI developers.
- The draft order threatens to withhold federal funds from states that violate guidelines related to AI regulations.
- The order responds to pressure from Big Tech advocates who claim that state laws create a confusing regulatory environment.
Artificial intelligence (AI) is a technology that allows machines to perform tasks that typically require human intelligence. There is ongoing debate over how to regulate AI, with many states enacting their own laws. This order, if signed, would align with a push for a unified federal approach to AI regulation.
Read the original ↗
Gemini 3 vs Grok 4.1: The Best AI of 2025 is…
The performance comparison between Gemini 3 and Grok 4.1 provides insights into the rapidly evolving landscape of AI models and their capabilities, influencing user choices across various applications.
- Google's Gemini 3 and xAI's Grok 4.1 were launched simultaneously, both claiming to be their best AI models yet.
- Gemini 3 focuses on improved reasoning, better memory, and deep multimodality for real-world usability.
- Grok 4.1 has shown significant enhancements in reasoning speed, factual accuracy, and reduced hallucination rates.
- Benchmark results indicate Grok 4.1 leads in some performance metrics, while Gemini 3 excels in image quality and detailed outputs.
Gemini 3 is Google's latest AI model with advanced capabilities, aiming to integrate seamlessly into its product ecosystem. Grok 4.1, developed by Elon Musk’s xAI, presents a competitive challenge with improvements over previous models. Both models signify the ongoing advancements in AI technology and set the stage for future innovations.
Read the original ↗
OpenAI Debuts GPT-5.1-Codex-Max, a Long-Horizon Agentic Coding Model With Compaction for...
The introduction of OpenAI's latest model signifies an advancement in coding capabilities, potentially enhancing productivity in software development and multi-window workflows.
- OpenAI has launched GPT-5.1-Codex-Max, a new model designed for coding tasks.
- The model features long-horizon agentic capabilities, allowing for extended task management in coding.
OpenAI is a leading artificial intelligence research organization known for developing language models. Previous iterations, such as GPT-3 and Codex, have advanced coding and programming tasks. The GPT-5.1-Codex-Max model builds on these foundations to further enhance coding productivity.
Based on the headline only — the full article could not be fetched.
Read the original ↗
The First "Generative AI Grill" Is Actually Great (No Thanks to AI)
The Zelos-450 grill demonstrates that appealing design and functionality can attract consumers, even if the generative AI features are not particularly effective.
- The Brisk It Zelos-450 is advertised as the first grill with generative AI, named Vera, for customized recipe creation.
- Despite the AI aspect being considered gimmicky, the grill is well-designed, easy to assemble, and priced under $300 during sales.
- The app for the Zelos-450 operates smoothly, allowing for remote temperature control and automatic ignition features not found in more expensive models.
Generative AI technology aims to personalize user experiences, but in the context of cooking, its utility is questioned. Grill technology has evolved significantly, with smart grills incorporating automated features to enhance user convenience. Consumers might benefit from focusing on overall grill performance and features rather than solely on AI enhancements.
Read the original ↗
Google DeepMind Hires Former CTO of Boston Dynamics as the Company Pushes Deeper Into Robotics
The hiring of a top robotics engineer reflects DeepMind's commitment to advancing AI technology in robotics, potentially shaping the future of robot functionality and integration.
- Google DeepMind has appointed Aaron Saunders, the former CTO of Boston Dynamics, as VP of hardware engineering.
- Saunders was responsible for developing advanced legged robots and joins DeepMind amid plans for an AI operating system called Gemini.
- DeepMind aims for Gemini to work seamlessly across different robot designs, much like Android does for smartphones.
Google DeepMind is a subsidiary of Alphabet focused on AI research. Boston Dynamics is renowned for its innovative robotics, particularly legged robots. The robotics field is rapidly evolving, with both U.S. and Chinese companies making significant advancements in humanoid and four-legged robots.
Read the original ↗
Nvidia’s record $57B revenue and upbeat forecast quiets AI bubble talk
Nvidia's substantial revenue growth signals strong demand for AI-related technologies, which may influence industry investment and development strategies moving forward.
- Nvidia reported a record revenue of $57 billion for the third quarter, up 62% year-over-year.
- The company's data center business generated $51.2 billion in revenue, representing a 66% increase from the previous year.
- Nvidia's forecast predicts fourth-quarter revenue will reach $65 billion, indicating continued growth in the AI sector.
- Geopolitical issues prevented Nvidia from shipping competitive products to China, affecting its potential revenue from the region.
- Sales of Nvidia's Blackwell GPU chips have soared, showcasing the rising demand for powerful computing solutions.
Nvidia is a leading technology company known for its graphics processing units (GPUs), which are essential for AI and gaming applications. This strong performance follows a trend of increased investment in AI technologies across various markets. Future developments to watch include Nvidia's product launches and government relations affecting international trade.
Read the original ↗
“We’re in an LLM bubble,” Hugging Face CEO says—but not an AI one
Understanding the distinction between large language models (LLMs) and broader AI applications is crucial for shaping future investment strategies in technology sectors.
- Clem Delangue, CEO of Hugging Face, claims there is a bubble specific to large language models, not AI as a whole.
- At an Axios event, Delangue suggested that the LLM bubble might burst next year, focusing investment on general-purpose chatbots.
- Delangue argues for a future with various specialized models rather than a one-size-fits-all approach to AI.
- Research firm Gartner noted a shift toward specialized models driven by business workflow needs and accuracy demands.
- Jeff Bezos co-founded a new AI startup for machine learning in engineering and manufacturing with over $6 billion in funding.
Hugging Face is a platform that provides resources for developing and sharing machine learning models, including specialized models beyond LLMs. The AI market is currently seeing over-concentration on LLMs, while investment in diverse AI applications is in its early stages. Key trends to watch include the development of specialized AI models and new funding initiatives in the sector.
Read the original ↗
An Implementation of a Comprehensive Empirical Framework for Benchmarking Reasoning Strategies in Modern Agentic AI Systems
The development of a comprehensive framework for benchmarking reasoning strategies in AI is crucial for assessing the effectiveness and reliability of modern AI systems.
- A comprehensive empirical framework for benchmarking reasoning strategies in agentic AI systems has been implemented.
- The framework aims to enhance the evaluation processes of reasoning techniques in contemporary AI.
Benchmarking is vital for understanding and improving AI performance, especially as systems become more sophisticated. This new framework could help standardize evaluations across various AI applications.
Based on the headline only — the full article could not be fetched.
Read the original ↗
VC Jennifer Neundorfer explains how founders can stand out in a crowded AI market
The insights from VC Jennifer Neundorfer highlight key strategies for founders to distinguish themselves in an increasingly competitive AI market, impacting their chances for successful funding and sustainable growth.
- Jennifer Neundorfer, co-founder of January Ventures, spoke about the challenges founders face in a saturated AI market during the Equity podcast at TechCrunch Disrupt.
- Neundorfer expressed a preference for founders creating entirely new experiences with AI, rather than those making only incremental improvements.
- She emphasized the importance of clearly communicating unique value propositions to investors to stand out among numerous similar startups.
January Ventures is a venture capital firm focused on investing in innovative technology startups, especially in the AI sector. As AI technology evolves rapidly, many founders struggle to present their ideas as unique due to market saturation. Moving forward, it is crucial for startups to adapt and innovate to survive potential market corrections.
Read the original ↗
Target Partners With OpenAI Through New ChatGPT Shopping App
The partnership between Target and OpenAI through a new shopping app could significantly enhance the online shopping experience by integrating AI capabilities to streamline product searches and purchases.
- OpenAI will launch a new ChatGPT-powered shopping app for Target, starting its beta rollout next week.
- The app will allow shoppers to search for products, build multi-item carts, and complete checkout processes directly in ChatGPT.
- Target will implement ChatGPT Enterprise for 18,000 employees to improve supply chain forecasting and in-store processes.
- The app will assist users in finding meal ideas, gifts, and navigating Target's product offerings through a chat interface.
OpenAI's ChatGPT technology enables conversational AI applications, and this collaboration with Target aims to enhance customer shopping experiences. OpenAI has recently partnered with other major retailers like Etsy and Shopify to expand its offerings. This move reflects an increasing trend in integrating AI into everyday shopping and retail operations.
Read the original ↗
Function Health raises $298M Series B at $2.5B valuation
The funding enables Function Health to enhance its AI capabilities, potentially transforming how individuals manage their health data and personal health information.
- Function Health raised $298 million in a Series B funding round led by Redpoint Ventures, achieving a valuation of $2.5 billion.
- The latest round increases Function's total capital raised to $350 million, with participation from various prominent venture firms and investors.
- Function Health aims to consolidate health data into a usable format through its Medical Intelligence Lab, which will develop an AI model for personalized health insights.
- The company offers an AI chatbot that provides tailored health guidance based on users' lab results and medical records, meeting HIPAA standards for data privacy.
- Function's advanced MRI service, Function Scans, is available at 130 locations and plans to expand to 200 by the end of 2025.
Function Health is a health technology company that specializes in consolidating various health data sources to enhance user health management. It previously focused on lab testing services and is now integrating artificial intelligence to offer personalized health insights. Future developments will include the expansion of their MRI service and further enhancements to their AI capabilities.
Read the original ↗
Critics scoff after Microsoft warns AI feature can infect machines and pilfer data
The potential security risks associated with Microsoft's new AI feature may expose users to significant vulnerabilities, emphasizing the need for better understanding and safeguards against data theft and device infections.
- Microsoft introduced Copilot Actions, an experimental AI feature for Windows that can perform tasks like organizing files and scheduling meetings.
- The company warned users about the risks of enabling Copilot Actions, stating it can infect devices and steal sensitive data.
- Critics argue that Microsoft’s warning reflects a larger pattern of rushing features to market without fully understanding their security implications.
- Security experts noted that similar warnings about other software features, like macros, have been ineffective in preventing exploitation.
- Copilot Actions is currently off by default, but critics fear it could eventually be made a standard feature without sufficient user awareness of risks.
Microsoft’s Copilot Actions aims to enhance productivity through AI but warns users of potential data security threats related to its features. Previous AI features have often transitioned from experimental to default, raising concerns about user vulnerability to attacks. Future developments need to balance innovation with robust security measures and user education.
Read the original ↗
The US Needs an Open Source AI Intervention to Beat China
The emergence of advanced open-source AI models from Chinese companies poses a competitive threat to the U.S. and highlights the need for strategic innovation in open-weight AI.
- The U.S. leads in AI innovation but risks falling behind in open-weight models essential for local adaptation and use.
- Chinese companies have rapidly gained popularity with open models, surpassing U.S. offerings in capability and developer support.
- Experts advocate for the U.S. to develop open models to mitigate reliance on foreign technology and foster innovation.
The U.S. AI landscape is dominated by companies like OpenAI and Google, which have released open models, but current offerings are less capable than those from China. The ATOM Project aims to address these concerns and promote open-source AI initiatives in the U.S. Moving forward, the U.S. may need to adopt more transparent and radical approaches to AI development and data sharing to maintain competitiveness.
Read the original ↗