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
OpenAI, Anthropic, and Block Are Teaming Up to Make AI Agents Play Nice
The establishment of the Agentic AI Foundation aims to create open standards, potentially transforming how AI agents communicate and operate across various applications, impacting businesses and consumers alike.
- OpenAI, Anthropic, and Block have founded the Agentic AI Foundation to promote standards for artificial intelligence agents.
- The foundation includes technologies like Anthropic’s Model Context Protocol and OpenAI’s Agents.md for building and connecting AI agents.
- Major companies including Google, Microsoft, and AWS have joined the foundation, indicating broad industry support for open-source AI standards.
The Agentic AI Foundation, established under the Linux Foundation, represents a shift towards developing software that enables AI agents to interact seamlessly. Historically, the AI industry has been dominated by proprietary models, but this initiative seeks to foster an ecosystem of open interoperability.
Read the original ↗
OpenAI Staffer Quits, Alleging Company’s Economic Research Is Drifting Into AI Advocacy
Concerns about OpenAI's research direction highlight potential biases in AI advocacy, which may affect public trust and understanding of the technology's economic impact.
- OpenAI's economic research team has seen employee departures amid claims of a cautious approach to publishing on AI's negative economic impacts.
- Former employee Tom Cunningham expressed difficulty in conducting rigorous analysis due to the team's advocacy role for OpenAI.
- OpenAI's chief strategy officer stated that the company must act as a responsible leader and not ignore problems associated with AI.
- Despite claims of expansive research efforts, sources allege a shift toward publishing more positive findings about AI, sidelining concerns like job displacement.
OpenAI, a prominent AI research organization, has been publishing studies since 2016 on the economic effects of its technologies. However, recent reports suggest a growing reluctance to release research indicating negative outcomes, raising questions about transparency. Observers are watching to see how this tension between advocacy and research quality evolves and influences public discourse.
Read the original ↗
OpenAI, Anthropic and Block join new Linux Foundation effort to standardize the AI agent era
Establishing standard protocols for AI agents can lead to improved interoperability and trust, addressing current concerns about fragmentation in the technology landscape.
- The Linux Foundation has launched the Agentic AI Foundation (AAIF) to standardize AI agents and avoid fragmentation.
- Anthropic, Block, and OpenAI have contributed tools to the AAIF, including the Model Context Protocol and Goose framework.
- The AAIF aims to create open standards that enable seamless integration among AI technologies without proprietary restrictions.
The Agentic AI Foundation is designed to promote open-source solutions for AI agents, addressing the growing complexity in AI interactions. This initiative builds on existing Linux Foundation projects but focuses specifically on AI agent interoperability. It is part of a broader push in the industry for collaborative development of AI technologies.
Read the original ↗
India proposes charging OpenAI, Google for training AI on copyrighted content
This proposal could alter the operational landscape for AI companies in India, potentially impacting how they compensate creators for the use of copyrighted content.
- India has proposed a royalty system requiring AI companies to pay for training on copyrighted content.
- The framework aims to ensure fair compensation for creators while reducing compliance costs for AI firms.
- AI companies are given a 30-day period to respond to the government's proposal.
India's Department for Promotion of Industry and Internal Trade has introduced this initiative in response to growing concerns over AI's use of copyrighted material. The proposal could significantly affect major players like OpenAI and Google, as India is a key market for AI technology.
Read the original ↗
Microsoft to invest $17.5B in India by 2029 as AI race accelerates
Microsoft's substantial investment in India highlights the country's growing significance in the global AI landscape and signifies a shift in tech competition among top firms.
- Microsoft plans to invest $17.5 billion in India from 2026 to 2029, which is the company's largest investment in Asia.
- The investment will be used to establish new data centers, AI infrastructure, and skilling programs to expand Microsoft's presence in the region.
- This initiative aligns with India's strategy to enhance digital infrastructure and AI adoption amid rising demand for technology services.
- Microsoft aims to train 20 million people in AI skills by 2030, building on its previous effort that trained 5.6 million since January.
- This expansion project includes the opening of a new data center in Hyderabad by mid-2026 and enhancements to existing facilities. The initiative is part of a broader trend of tech companies increasing investments in India's AI capabilities.
Microsoft, a major tech corporation, is expanding its investments in India due to the country's large internet user base and growing opportunities in AI and cloud services. This announcement follows other significant commitments from global companies, indicating India's emerging role as a strategic location for tech expertise and cloud infrastructure. Moving forward, expect increased competition among tech giants in India as they enhance their service offerings.
Read the original ↗
EU launches antitrust probe into Google’s AI search tools
This investigation may reshape the landscape of competition in the AI market, particularly regarding content usage and compensation associated with AI-generated search results.
- The European Commission has launched an antitrust investigation into Google for potential competition law violations.
- The probe will assess if Google is using website content without proper compensation to produce AI summaries above search results.
- The investigation will examine how Google's AI tools utilize videos from YouTube in generating responses.
- Google's practices could limit competition by not allowing rivals to use YouTube content for AI training, raising concerns among regulators.
Google's AI Overview and AI Mode are central to the investigation, focusing on their use of web publishers' content. This inquiry occurs amidst broader discussions about AI regulation in the EU and ongoing lawsuits by publishers against various AI companies for copyright infringement. The EU is trying to ensure fair competition while simplifying its AI regulatory framework.
Read the original ↗
Google’s first AI glasses expected next year
The development of AI-powered glasses by Google signifies a competitive expansion in wearable technology, highlighting the evolution of consumer electronics in everyday life.
- Google will launch its first AI glasses in 2026, featuring various models for different user needs.
- One model offers screen-free assistance with built-in speakers, microphones, and cameras for communication and photography.
- Another model includes an in-lens display for navigation and closed captioning, viewable only to the wearer.
Google is entering the smart glasses market amidst rising competition from companies like Meta, Apple, and Snap. Previous developments include partnerships with eyewear companies to enhance accessibility and functionality in smart eyewear. Future updates will likely focus on consumer adaptation and user experience enhancements in this emerging technology sector.
Read the original ↗
UnityVideo: Unified Multi-Modal Multi-Task Learning for Enhancing World-Aware Video Generation
UnityVideo's innovations in multi-modal learning could significantly advance the field of video generation, enhancing applications in entertainment, education, and other industries through better quality and understanding of real-world contexts.
- UnityVideo is a framework designed to enhance video generation by integrating various modalities such as segmentation masks, human skeletons, and depth maps.
- The approach employs dynamic noising for unifying training paradigms and a modality switcher for improved processing.
- A large-scale dataset featuring 1.3 million samples supports the framework, aiming to facilitate better video quality and generalization to unseen data.
UnityVideo represents a significant advancement in video generation technologies, addressing limitations of existing single-modality models by enhancing cross-modal interactions. By improving video alignment with physical constraints, this work is set to influence various applications in artificial intelligence and computational media. Future developments may focus on refining these multi-modal capabilities and expanding applications in real-world scenarios.
Read the original ↗
ReCamDriving: LiDAR-Free Camera-Controlled Novel Trajectory Video Generation
ReCamDriving's advancements in video generation technology may enhance the creation of realistic and dynamic visual content for a range of applications in entertainment and simulation.
- ReCamDriving is a camera-controlled video generation framework that uses dense 3D geometric scene (3DGS) renderings for precise control.
- The framework employs a two-stage training approach that first uses camera poses for coarse control, followed by fine-grained guidance with 3DGS renderings.
- It introduces a 3DGS-based cross-trajectory data curation strategy to improve camera transformation patterns in training, resulting in the ParaDrive dataset with over 110,000 video pairs.
ReCamDriving represents a shift from traditional video generation methods, utilizing complete scene representations for enhanced realism. This approach addresses challenges faced by previous methods relying on sparse data. Future developments may focus on expanding applications and improving processing efficiency.
Read the original ↗
Mistral AI surfs vibe coding tailwinds with new coding models
The release of Mistral's new AI coding models may enhance coding productivity and competition in the rapidly evolving AI development landscape.
- Mistral AI launched Devstral 2, a new coding AI model, to compete with larger firms like Anthropic.
- The new Devstral 2 model features 123 billion parameters and requires significant computing resources for deployment.
- Mistral Vibe, a new command-line interface, aims to simplify code automation through natural language processing and context awareness.
- Devstral 2 is currently free via API, with future costs estimated at $0.40/$2.00 per million tokens for usage.,
Mistral AI is a French startup specializing in artificial intelligence, focusing on developing advanced AI models for various practical applications. This launch signifies Mistral's strategic efforts to position itself among leading AI developers, especially in coding-related tools. Future developments to watch include updates on the performance and adoption of Devstral models in commercial settings.
Read the original ↗
Pebble’s founder introduces a $75 AI smart ring for recording brief notes with a press of a button
The Index 01 smart ring offers a new, privacy-preserving solution for quickly capturing ideas without the constant surveillance of traditional AI devices.
- The Index 01 smart ring is priced at $75 and allows users to record brief notes and reminders by pressing a button.
- The device uses open-source, local AI technology that operates through a smartphone app, recording only when the button is pressed.
- Unlike fitness trackers, the ring is designed solely for capturing thoughts and does not track health metrics or use a subscription model.
Core Devices, founded by Eric Migicovsky, is expanding its lineup with the Index 01 smart ring, which offers a unique approach to quick note-taking. This product is part of a growing trend in wearable technology aimed at simplifying personal data capture without compromising user privacy.
Read the original ↗
Anthropic and Accenture sign multi-year AI strategic partnership
This partnership aims to enhance AI capabilities for Accenture's workforce, potentially improving productivity and AI implementation across various sectors.
- Anthropic has entered into a multi-year partnership with Accenture, confirmed by both companies to last three years.
- The collaboration will form the Accenture Anthropic Business Group, providing formal AI training to Accenture's 30,000 employees.
- Anthropic's Claude coding tools will be accessible to Accenture's developers, supporting AI development efforts.
Anthropic is an AI research lab focusing on enterprise solutions and has recently gained significant market share, holding 40% of the enterprise market. This partnership follows earlier collaborations with other firms, indicating Anthropic's expanding influence in the AI sector.
Read the original ↗
Your Brain Rewrites Your Memories Each Time You Recall Them, New Study Finds
This research challenges the traditional view of memory as static, highlighting the importance of understanding memory's malleability for various fields, including psychology and law.
- A neuroscience review reveals that memories are reconstructed each time they are recalled, altering their accuracy and clarity.
- The researchers reviewed nearly 200 studies to better define memory representations and how they change with each recollection.
- They propose that the hippocampus, a key brain structure for memory, allows memory traces to remain inactive until activated by a cue.
The study, published in Neuroscience & Biobehavioral Reviews, draws on findings from psychology, neuroscience, and philosophy to explore the dynamics of episodic memory. This research provides insights into how memories can be distorted by time, context, and our own perceptions.
Read the original ↗
iFixit Put a Chatbot Repair Expert in an App
The introduction of the FixBot chatbot app by iFixit enhances accessibility to repair information, potentially empowering users to fix their devices independently and prolong their lifespan.
- iFixit has released a new chatbot app, FixBot, for iOS and Android that assists users with device repairs.
- The app provides health checks for devices and connects users to over 125,000 repair guides in iFixit's library.
- Users can interact with FixBot using voice or text and utilize their camera to identify issues with devices for tailored repair assistance.
iFixit is an online repair service known for its comprehensive library of repair manuals and guides. FixBot improves user experience by employing AI technology to streamline the repair process, facilitating easier access to troubleshooting and replacement parts.
Read the original ↗
Empromptu raises $2M pre-seed to help enterprises build AI apps
Empromptu's funding and platform development will enable businesses, even those without technical skills, to create AI applications, potentially transforming enterprise software management.
- Empromptu raised $2 million in a pre-seed funding round led by Precursor Ventures.
- The platform allows users to instruct an AI chatbot to build applications such as classification or recommendation tools.
- New features announced include custom data models and infinite memory, enhancing the platform's capabilities.
Empromptu is a startup co-founded by Shanea Leven and Sean Robinson, designed to help enterprises build AI applications without requiring extensive technical knowledge. Following its launch last October, the company is focusing on industries with regulatory complexities and has recently secured funding to expand its technology and team.
Read the original ↗
Anthropic Brings Claude Code Into Slack
The integration of Claude Code into Slack enhances coding assistance by embedding AI within team communication, potentially transforming how developers collaborate on coding tasks.
- Anthropic has introduced a beta version of Claude Code in Slack for developers to manage coding tasks within chat threads.
- Developers can tag @Claude to initiate full coding sessions based on ongoing conversations, such as bug reports and feature requests.
- This development aims to reduce workflow friction by eliminating the need for window-switching and minimizing additional documentation searches.
Claude is an AI coding assistant developed by Anthropic. Previously limited to integrated development environments (IDEs), coding assistants are now entering collaborative spaces like Slack, which could streamline coding processes and improve team interactions.
Read the original ↗
Unified Video Editing with Temporal Reasoner
The development of VideoCoF enhances video editing capabilities by achieving precision without needing user-provided masks, potentially streamlining workflows in video production.
- VideoCoF is a new video editing method that improves precision and instruction-to-region mapping without relying on user-provided masks.
- It employs a Chain-of-Frames approach which includes a reasoning step called 'see, reason, then edit' for better results.
- VideoCoF demonstrated state-of-the-art performance with only 50,000 training video pairs, indicating high efficiency and effectiveness.
Video editing traditionally relies on user-provided masks for accuracy, limiting broader applications. VideoCoF aims to unify these methods by improving spatial cue alignment through an innovative training approach. The success of VideoCoF could influence future video editing technologies and ease content creation processes.
Read the original ↗
TwinFlow: Realizing One-step Generation on Large Models with Self-adversarial Flows
The introduction of TwinFlow could significantly enhance the efficiency of generative models, impacting various applications in multi-modal generation such as text-to-image tasks.
- TwinFlow is a one-step generative model framework that improves inference efficiency without needing fixed pretrained teacher models or traditional adversarial networks.
- The model achieves a GenEval score of 0.83 in 1-number of function evaluations (NFE), outperforming existing frameworks like SANA-Sprint and RCGM.
- TwinFlow demonstrates scalability by enabling full-parameter training on Qwen-Image-20B and converting it into an efficient generator, reducing computational costs by 100 times with minor quality loss.
TwinFlow represents a new approach in the field of generative AI, particularly in accelerating inference for multi-modal tasks. Unlike traditional multi-step methods, TwinFlow focuses on efficiency and simplicity, without the overhead of auxiliary networks. Future developments related to TwinFlow could further enhance generative performance across various applications.
Read the original ↗
America’s Biggest Bitcoin Miners Are Pivoting to AI
The shift of bitcoin miners to AI reflects significant market changes, highlighting the transformation of data centers and the competitive landscape in computing technology.
- Many industrial-scale bitcoin mining companies are repurposing facilities for AI and high-performance computing due to declining bitcoin profitability.
- At least eight publicly traded bitcoin mining firms have announced plans to transition to AI, prompted by strong demand for AI data centers.
- In recent months, publicly traded bitcoin miners secured over $43 billion in contracts related to AI and HPC, revealing a lucrative market shift.
Bitcoin mining, which involves solving computational puzzles for transaction rewards, has seen profitability decline as competition increases and bitcoin prices drop. The rising demand for AI capabilities is driving miners to adjust their business models, moving away from traditional bitcoin mining to AI services.
Read the original ↗
OpenAI Should Stop Naming Its Creations After Products That Already Exist
Trademark disputes like the ones faced by OpenAI can significantly impact brand identity and product launches, raising questions about naming practices in the tech industry.
- OpenAI's app Sora allows users to generate digital likenesses for personalized deepfake videos, but faced a trademark lawsuit over its feature named 'cameo'.
- A US District Judge ordered OpenAI to temporarily stop using the name 'cameo' due to conflicts with the existing app Cameo, leading OpenAI to rebrand the feature as 'characters'.
- The legal conflict also involves OpenAI being barred from using the name 'io' for upcoming hardware because of a trademark claim from another company, iyO.
- The next court hearing concerning the ban on the name 'cameo' is set for December 19, which could affect OpenAI's branding strategy.
OpenAI is a prominent artificial intelligence company known for its generative AI technologies. In September, it launched Sora, an app for sharing AI-generated videos, which quickly became popular on iOS. Legal challenges related to branding may influence how the company names future products.
Read the original ↗