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
Humanoids, AVs, and what’s next in AI hardware at TechCrunch Disrupt 2025
The upcoming TechCrunch Disrupt 2025 will feature key discussions on AI hardware advancements that could significantly impact future robotic and autonomous systems development.
- TechCrunch Disrupt 2025 will occur from October 27 to 29 in San Francisco, attracting over 10,000 leaders from startups and venture capitalists.
- Raquel Urtasun from Waabi and Jeff Cardenas from Apptronik will discuss robotics and autonomous systems on the AI Stage.
- The session will focus on the complex systems required for autonomous vehicles and humanoid robots, including simulation and software infrastructure.
TechCrunch Disrupt is an annual conference that highlights innovation in technology, particularly in AI and robotics. Waabi and Apptronik are notable companies working on advanced robotics and autonomous vehicle technologies. Future developments in this field will likely address current challenges and unlock new capabilities.
Read the original ↗
Paper2Agent: Reimagining Research Papers As Interactive and Reliable AI Agents
Paper2Agent aims to enhance research paper accessibility and utility by transforming them into interactive AI agents, fostering easier knowledge sharing and collaborative analysis in the scientific community.
- Paper2Agent is an automated framework that converts research papers into AI agents, enabling complex scientific inquiries through natural language.
- The system leverages a Model Context Protocol (MCP) server to analyze papers and generate tests, ensuring the reliability of the resulting AI agents.
- Case studies demonstrate that these paper agents can reproduce original results and effectively address new user queries.
- Agents created through Paper2Agent utilize tools from their corresponding papers, fostering dynamic interaction with research findings.
Paper2Agent seeks to overcome barriers typically associated with traditional research papers, which require significant reader effort for analysis and application. By converting static papers into interactive agents, this initiative reshapes how research is disseminated and utilized. Future developments may include broader applications and enhancements in AI-driven research collaboration.
Read the original ↗
Spotify peeved after 10,000 users sold data to build AI tools
The situation highlights ongoing tensions between user data ownership rights and corporate control over personal information, raising questions about privacy rights in digital platforms.
- Over 10,000 Spotify users sold data via a collective called 'Unwrapped' to developers building AI tools, in violation of Spotify's developer policy.
- Spotify sent a warning to Unwrapped for potential trademark infringement and data policy violations, but Unwrapped claims it did not receive communication from Spotify.
- Spotify's terms prohibit using their content for AI model building, and the platform is reportedly making it difficult for users to port their data to Unwrapped.
Spotify is a music streaming service known for its annual 'Wrapped' feature that summarizes user listening habits. The collective Unwrapped aims to help users monetize their data by pooling it for sale to developers. Ongoing disputes about user data rights against corporate policies and data management practices are expected to continue.
Read the original ↗
Microsoft ends OpenAI exclusivity in Office, adds rival Anthropic
Microsoft's decision to add Anthropic's AI technology to Office 365 alongside OpenAI's reflects a strategic shift in the competitive AI landscape, promoting innovation and choice for users.
- Microsoft will incorporate AI models from Anthropic into its Office 365 suite, ending OpenAI's exclusivity for generative AI features.
- Anthropic's Claude Sonnet 4 model has reportedly outperformed OpenAI's models in tasks such as visual design and spreadsheet automation during internal testing.
- Microsoft will acquire access to Anthropic's models through Amazon Web Services, marking a unique collaboration between competing companies.
Microsoft has relied on OpenAI since their partnership began in 2019, investing over $13 billion to integrate AI features into its products. The addition of Anthropic comes amid a broader trend of technological diversification in AI, as both companies seek to expand their offerings.
Read the original ↗
After selling to Spotify, Anchor’s co-founders are back with Oboe, an AI-powered app for learning
Oboe's innovative approach to personalized learning through AI could significantly change how users engage with educational content and tailor learning experiences to their individual needs.
- Oboe is an AI-powered app allowing users to create custom learning courses on various topics by simply entering a prompt.
- The app offers multiple formats, including text, visuals, audio lectures, and interactive tests to cater to different learning preferences.
- At launch, users can take any course for free and create five courses per month, with paid tiers available for additional access.
- The co-founders, Nir Zicherman and Michael Mignano, previously launched Anchor, which was acquired by Spotify in 2020.
- Oboe raised $4 million in seed funding led by Eniac Ventures, with contributions from multiple investors.
Oboe is a newly launched educational app developed by the co-founders of Anchor after their exit from Spotify. It utilizes AI technology to create personalized learning experiences quickly. Future developments include expanding app availability beyond the web to mobile platforms.
Read the original ↗
Exploring the future of voice AI with Mati Staniszewski at TechCrunch Disrupt 2025
The advancements in voice AI technology could transform how humans interact with digital systems across various sectors, impacting accessibility and engagement.
- Mati Staniszewski, CEO of ElevenLabs, will discuss voice AI at TechCrunch Disrupt 2025, which occurs from October 27–29 in San Francisco.
- ElevenLabs specializes in generative AI, developing technology that can accurately replicate natural speech.
- The session will cover the technical and ethical challenges of creating human-like synthetic voices and their applications in entertainment and education.
ElevenLabs is a growing leader in AI-driven synthetic voice technology, recognized for enhancing the realism of AI-generated speech. Previous discussions around AI voice focused on technical limitations and ethical considerations, driving interest in its potential uses.
Read the original ↗
Machine Learning Tests Keep Getting Bigger
The introduction of new machine learning benchmarks reflects the rapid advancements in AI technology, impacting industry performance standards and computational requirements.
- Nvidia's Blackwell Ultra GPU topped MLPerf's new reasoning benchmark with significant performance improvements.
- MLPerf introduced three new benchmark tests, including the largest language model based on the DeepSeek-R1 671B model.
- AMD and Intel also participated, with AMD launching MI325X GPUs and Intel submitting its Arc Pro GPU for the first time.
MLPerf is a biannual competition that evaluates machine learning performance. Recent rounds have seen the introduction of increasingly complex models, emphasizing the industry's growth and demand for faster processing and accuracy. Future rounds will likely feature even larger models and diverse benchmarks.
Read the original ↗
Harvard AI Model Promises Breakthrough in Drug Discovery
The development of PDGrapher could significantly accelerate drug discovery and personalized treatment approaches, potentially transforming biomedical research and improving outcomes for patients with complex diseases.
- Researchers at Harvard Medical School created an AI model called PDGrapher, which aims to enhance drug discovery methods.
- PDGrapher uses a graph neural network to analyze multiple disease drivers and predict effective gene or drug combinations.
- The model demonstrated up to 35% higher accuracy and 25 times faster results compared to conventional models, showing promise in cancer treatment research.
- It is currently being applied to study various diseases, including cancer and neurodegenerative disorders such as Alzheimer's and Parkinson's.
- The tool is freely available and aims to provide personalized treatment recommendations based on individual patient profiles.
PDGrapher represents a shift from traditional drug discovery methods that test one compound at a time to a more comprehensive analysis of disease mechanisms. This model may lead to faster identification of effective treatments and personalized medicine applications. Ongoing research will explore its utility across various diseases and its impact on future drug development.
Read the original ↗
RSS co-creator launches new protocol for AI data licensing
The launch of the Real Simple Licensing (RSL) protocol could drastically change how AI companies acquire training data, potentially reducing legal risks in the industry.
- The Real Simple Licensing (RSL) system allows AI companies to license training data at scale, supported by major publishers including Reddit, Quora, and Yahoo.
- RSL addresses the legal issues surrounding unlicensed data usage, especially following Anthropic's $1.5 billion copyright settlement.
- The RSL Protocol specifies licensing terms that publishers set, making it easier to identify data usage rights through 'robots.txt' files.
- An RSL Collective has been formed to negotiate terms and collect royalties, similar to agreements in music and film industries.
Real Simple Licensing (RSL) is a new system designed for AI companies to license training data amid growing legal challenges regarding data usage. Available publishers have already joined the collective, but questions remain about major AI labs' willingness to participate in the licensing system. Future developments will likely hinge on the industry's adoption of these licensing terms as legal scrutiny intensifies.
Read the original ↗
Pay-per-output? AI firms blindsided by beefed up robots.txt instructions.
The introduction of the Really Simple Licensing standard could reshape the relationship between AI companies and content creators, potentially leading to fairer compensation for the use of online content in AI training.
- The Really Simple Licensing (RSL) standard adds an automated licensing layer to robots.txt files to ensure compensation for content used by AI crawlers.
- The RSL standard allows publishers to implement various compensation models, including pay-per-crawl and pay-per-inference.
- Developed by the RSL Collective, the standard aims to provide a scalable solution for AI companies to access content while compensating creators.
- The adoption of RSL could help preserve the revenue of publishers as AI tools increasingly reference their content without permission.
The RSL standard was created to give publishers control over the use of their content by AI firms, encouraging fair compensation in an era where AI training relies heavily on online sources. Previously, there was minimal guidance on licensing content for AI, resulting in legal disputes and concerns over content theft. Stakeholders now anticipate how major AI firms will respond to this new licensing framework as it becomes widely available.
Read the original ↗
AI vs. MAGA: Populists alarmed by Trump’s embrace of AI, Big Tech
This story highlights a growing divide within the Republican Party over the implications of artificial intelligence, which may affect political strategies leading into upcoming elections.
- Melania Trump praised artificial intelligence, calling it a potential 'great engine of progress' at a recent White House event with tech executives.
- Senator Josh Hawley warned that AI could threaten jobs and the liberty of the common man during a speech at a MAGA conference.
- A rising faction within the MAGA movement is increasingly critical of Big Tech, despite Trump’s administration promoting AI adoption.
- Polling indicates significant skepticism about AI among voters, with more than 55 percent objecting to a proposed moratorium on state regulations of AI.
- Some Republicans are calling for restraint on AI's growth to prevent job displacement and social consequences, viewing it as a threat to societal stability.
The embrace of AI by the Trump administration is met with skepticism from a growing segment of the MAGA movement, who fear its impact on jobs and societal values. This reflects a complex relationship between Big Tech and conservative ideologies within the party, highlighting potential challenges ahead. Observers are watching how this internal conflict may influence the Republican Party's strategies as the midterm elections approach.
Read the original ↗
DARPA’s New “MASH” Program Aims to Deploy Autonomous Robot Medics to Save Lives on the Battlefield
The MASH program could transform battlefield medicine, potentially saving lives by providing immediate care for severe injuries that traditional methods cannot treat effectively during combat.
- DARPA has launched the MASH program, focusing on robotic systems to treat non-compressible torso hemorrhage on the battlefield.
- The program aims to stabilize such injuries for over 48 hours until soldiers can reach advanced surgical care.
- Teams of experts in trauma surgery, robotics, and AI will collaborate over 36 months to develop these autonomous medical solutions.
MASH, short for Medics Autonomously Stopping Hemorrhage, is a new initiative from DARPA to address critical battlefield injuries. Previous advances in military robotics have laid the groundwork for potential breakthroughs in medical intervention.
Read the original ↗
AI gaming startup Born raises $15M to build ‘social’ AI companions that combat loneliness
The rise of social AI companions like Born's could provide innovative solutions to alleviate feelings of loneliness in an increasingly digital world.
- AI gaming startup Born has raised $15 million in Series A funding, bringing its total funding to $25 million.
- Born's app features a virtual pet named Pengu, designed to promote social collaboration among users, enhancing real-world relationships.
- The startup plans to expand its offerings with new characters for the Pengu app and a separate AI social product for users aged 16 to 21.
Born, formerly known as Slay, aims to improve digital interactions by creating emotionally intelligent AI companions. The company focuses on developing entertainment that fosters shared experiences and social bonds. Future developments include launching culturally relevant AI characters that enhance user engagement alongside existing apps.
Read the original ↗
AI Data Centers’ Water Use Is a Growing Concern
Growing water consumption due to AI data centers poses environmental challenges, impacting local water supplies and complicating infrastructure decisions in water-stressed regions.
- Data centers supporting AI consume significant amounts of water, with a 2023 estimate of 17.5 billion gallons used directly in the U.S.
- Electricity generation for data centers also contributes to water use, estimated to account for over 80% of total consumption.
- Recent projections suggest direct water consumption by U.S. data centers could double or quadruple by 2028, stressing local water systems.
AI data centers utilize cooling systems that often rely on extensive water use, raising concerns over their environmental impact in areas facing water shortages. The growing demand for AI services leads to increased strain on local water supplies, necessitating innovative approaches for sustainable cooling solutions.
Read the original ↗
Nvidia Introduces Rubin CPX GPU for Long-Context AI
The introduction of the Rubin CPX GPU represents a significant advancement in handling large data contexts, enabling more effective AI applications in video generation and long-form content creation.
- Nvidia has announced the Rubin CPX GPU at the AI Infrastructure Summit, designed for long-context tasks.
- The CPX can handle context windows larger than one million tokens, improving AI models' ability to process larger sequences.
- This GPU supports Nvidia's 'disaggregated inference' approach, enhancing scalability and reducing bottlenecks in large AI applications.
- The Rubin CPX is expected to launch at the end of 2026, reflecting Nvidia's commitment to innovation in AI technology.
Nvidia is a leading technology company known for its graphics processing units (GPUs) and AI solutions. The Rubin series aims to address limitations of existing AI models in handling extensive data sequences. Analysts will be watching how this GPU impacts the market and its application across various AI tasks moving forward.
Read the original ↗
iPhone 17, iPhone Air, AirPods Pro 3, and everything else announced at Apple’s hardware event
The announcement of new iPhone models and hardware updates signifies Apple's ongoing innovation in a competitive smartphone market, impacting consumer choices and industry standards.
- Apple unveiled the iPhone 17, which features a larger 6.3-inch screen and a new 48-megapixel ultrawide camera.
- The new ultra-thin iPhone Air, at just 5.6 mm thick, replaces the iPhone Plus and supports eSIM only.
- Apple also introduced updated Apple Watch models with blood pressure monitoring capabilities and new AirPods Pro 3 with live translation technology.
- The iPhone 17 starts at $799, while the iPhone Air is priced at $999 and the Apple Watch Series 11 is priced at $399.
- Future products such as a new iPad Pro and updated MacBooks are expected in 2025 or early 2026.
Apple is a leading technology company known for its smartphones, computers, and other electronic devices. This event follows previous hardware releases where Apple introduced significant updates. Observers will be watching for announcements regarding future technology and products, especially any relating to AI enhancements.
Read the original ↗
Ex-Google X trio wants their AI to be your second brain — and they just raised $6M to make it happen
The development of TwinMind represents a significant advance in AI applications that aim to enhance personal productivity and information retention in everyday life.
- TwinMind, co-founded by three former Google X scientists, has raised $5.7 million in seed funding for its AI-powered app that passively captures ambient speech.
- The app runs on Android and iOS, converting spoken thoughts into structured memory for generating notes, to-dos, and answers, without sending recordings to the cloud.
- It supports real-time translation in over 100 languages and can capture audio continuously for 16 to 17 hours without draining battery.
- TwinMind has over 30,000 users, with a significant portion utilizing the app for professional and personal purposes, and it also differentiates itself by working offline.
TwinMind is an AI application designed to enhance users' productivity by capturing and processing speech through passive audio recording. The app emerged from the experiences of its co-founders at Google X and aims to provide an alternative to traditional note-taking methods. Future developments may include enhanced features and user acquisition strategies as the startup continues to grow.
Read the original ↗
Google’s former security leads raise $13M to fight email threats before they reach you
The launch of AegisAI and its real-time AI agents aims to significantly reduce email threats, addressing a crucial aspect of cybersecurity amid rising phishing attacks.
- AegisAI, founded by former Google security leads, has secured $13 million in seed funding to combat email threats using AI agents.
- Over 90% of successful cyberattacks start with phishing emails, highlighting the urgent need for effective email security solutions.
- The startup offers a network of autonomous AI agents that analyze and neutralize email threats without relying on static rules, reducing false positives by up to 90%.
- Installation of AegisAI’s system on Google Workspace or Microsoft 365 takes about five minutes, with a pilot program already active in the U.S. and Europe.
AegisAI is a new email security startup utilizing AI to autonomously neutralize threats such as phishing and malware. It was founded by experienced professionals from Google's cybersecurity teams and is currently in its initial rollout phase with corporate clients.
Read the original ↗
MobiAgent: A Systematic Framework for Customizable Mobile Agents
The development of MobiAgent represents a significant leap in mobile agent technology, enabling greater cost-efficiency in data collection and better performance in practical applications.
- MobiAgent is a comprehensive mobile agent system that includes MobiMind models, AgentRR framework, and MobiFlow benchmarking suite.
- The system achieves state-of-the-art performance in real-world mobile scenarios, outpacing models like GPT-5 and Gemini-2.5-pro.
- MobiAgent features an AI-assisted agile data collection pipeline that reduces manual annotation costs.
- The entire system is open-sourced, allowing for widespread accessibility and potential further development.
- MobiAgent's deployment utilizes Ascend 910B NPU technology.
MobiAgent is designed to enhance mobile agent capabilities amidst challenges like accuracy and efficiency in real-world tasks. It follows the growing trend of Vision-Language Models that integrate visual and textual information. Future updates may include broader language support and additional features for various applications.
Read the original ↗
Open Deep Search: Democratizing Search with Open-source Reasoning Agents
Open Deep Search aims to enhance open-source AI capabilities, potentially reducing reliance on proprietary systems and promoting broader accessibility and innovation in AI search applications.
- Open Deep Search integrates a reasoning agent and a search tool with open-source language models to improve their functionality.
- The system uses two main components: the Open Search Tool and the Open Reasoning Agent, which enhance task execution with web searches.
- ODS improves accuracy on benchmarks SimpleQA and FRAMES, surpassing proprietary models like GPT-4o Search Preview by 9.7% on FRAMES.
Open Deep Search is an initiative designed to make advanced search functionalities accessible through open-source language models. This follows a trend where proprietary AI solutions dominate the search market, highlighting a need for competitive alternatives. Key systems to watch include OpenAI's GPT-4o and Perplexity's Sonar Reasoning Pro.
Read the original ↗