Latest AI News
The Forces Pushing Automobile GCCs in India Towards Digital Reinvention
Nearly 20% of these centres have been established in just the last three years, reflecting strong industry momentum.
View
Andrew Ng Builds a Stack Overflow for AI Coding Agents
Chub is designed for your coding agent to use (not for you to use!).”
View
Yotta, Gorilla Technologies Sign AI Infrastructure Deal Expected to Generate $500 Mn
Under the agreement, Yotta Data Services will operate 640 high-performance servers with more than 5,000 GPUs for AI workloads.
View
TCS Launches Rapid Outcome AI Platform With NVIDIA to Fast-Track Enterprise AI Adoption
The new platform blends generative, predictive and vision AI to move enterprises from pilots to production at scale.
View
Adobe, NVIDIA Partner to Develop Next-Gen Firefly Models for Creative Marketing
Adobe and NVIDIA will focus on developing next-generation Adobe Firefly models.
View
LTM Ties Up With IIT Kharagpur for Demand-Led AI Skilling of Workforce
Partnership combines training and research to build project-aligned AI capabilities across teams.
View
Memories AI is building the visual memory layer for wearables and robotics
Shawn Shen believes that AI will need to remember what it sees in order to succeed in the physical world. Shen’s companyMemories.aiis using Nvidia AI tools to build the infrastructure for wearables and robotics to be able to remember and recall visual memories. Memories.ai announced a collaboration with semiconductor giant Nvidia at its GTC conference on Monday. Through this partnership, Memories.ai uses Nvidia’s Cosmos-Reason 2, a reasoning vision language model, and Nvidia Metropolis, an application for video search and summarization, to continue to develop its visual memory technology. Shen (pictured above left) told TechCrunch that he and his co-founder and CTO, Ben Zhou (pictured above right), got the idea for the company while building the AI system behind Meta’s Ray-Ban glasses. Building the AI glasses got them thinking about how people would actually use the tech in real life if users couldn’t recall the video data they were recording. They looked around to see if they could find anyone already building that type of visual memory solution for AI. When they couldn’t, they decided to spin out of Meta and build it themselves. “AI is already doing really well in the digital world. What about the physical world?” Shen said. “AI wearables, robotics need memories as well. … Ultimately, you need AI to have visual memories. We believe in that future.” The ability for AI systems to remember, in general, is relatively new.OpenAI updated ChatGPT to start to remember past chats in 2024andfine-tuned that feature in 2025.Elon Musk’s xAIandGoogle Geminihave also launched their own memory tools in the past two years. But these advancements have largely focused on text-based memory, Shen said. Text-based memory is much more structured and easier to index but isn’t as helpful for physical AI applications that largely interact with the world through sight and visuals. Memories.ai was launched in 2024 and has raised $16 million thus far, through an $8 million seed round in July 2025 and an $8 million extension. The round was led by Susa Ventures and included Seedcamp, Fusion Fund, and Crane Venture Partners, among others. Shen said successfully building this visual memory layer required two things: building the infrastructure needed to embed and index videos into a data format that can be stored and recalled, and capturing the data needed to train the model to do just that. The company launched itslarge visual memory model (LVMM) in July 2025. Shen said it could be compared to a smaller version ofGemini Embedding 2, a multimodal indexing and retrieving model, that was released earlier this month. For data collection, the company created LUCI, a hardware device worn by the company’s “data collectors” that records video used to train the model. Shen said they don’t plan to become a hardware company, nor sell these devices, but, rather, that they built their own because they weren’t satisfied with off-the-shelf video recorders that focused on high-definition and battery-eating video formats. The company released the second generation of this LVMM and signed apartnership with Qualcommto run on Qualcomm’s processors starting later this year. Memories.ai is also working with some of the large wearable companies already, Shen said, but declined to disclose which ones. Despite some demand now, Shen sees even bigger opportunities in wearables and robotics yet to come. “In terms of commercialization, we are more focused on the model and the infrastructure, because ultimately we think the wearables and robotics market will come, but it’s probably just not now,” Shen said.
View
Warren presses Pentagon over decision to grant xAI access to classified networks
Sen. Elizabeth Warren (D-MA) sent aletterto Defense Secretary Pete Hegseth on Monday expressing concern over the Pentagon’s decision to give Elon Musk’s company xAI access to classified networks. “Grok, the controversial AI model developed by xAI, has provided disturbing outputs for users, including giving users ‘advice on how to commit murders and terrorist attacks,’ generating antisemitic content, and creating child sexual abuse material,” the letter reads. Warren said Grok’s “apparent lack of adequate guardrails” could pose “serious risks to the safety of U.S. military personnel and to the cybersecurity of classified systems.” She demanded Hegseth provide information on how the Department of Defense plans to “mitigate these potential national security risks.” Warren isn’t the first to express alarm at Grok, xAI’s controversial chatbot, gaining access to classified systems. Last month, acoalition of nonprofits urgedthe government to immediately suspend the deployment of Grok infederal agencies, including the DoD, after X users repeatedly prompted the chatbot to turn real photos of women, and in some cases children, intosexualized imageswithout their consent. The same day Warren sent her letter, a class action lawsuit was filed against xAI alleging Grok had generated sexual content from real images of the plaintiffs as minors. The letter comes in the aftermath of the Pentagon’s decision tolabel Anthropic a supply chain riskafter the AI firm refused to give the military unrestricted access to its AI systems. Anthropic had been, until recently, the only AI company with classified-ready systems. In the midst of that conflict, the DoD signed anagreement with OpenAIas well as xAI to use the two companies’ AI systems in classified networks, according toAxios. A senior Pentagon official confirmed that Grok was onboarded to be used in a classified setting, but is not yet being used. “It is unclear what assurances or documentation xAI has provided to the Department of Defense about Grok’s security safeguards, data-handling practices, or safety controls, and whether DoD has evaluated those assurances before reportedly allowing Grok access to classified system,” Warren writes. Warren requested a copy of the deal reportedly reached between the DoD and xAI on the use of Grok in classified systems and an explanation of how the department plans to ensure Grok is not exposed to cyberattacks and will “not leak sensitive or classified military information.” (Last week, a former employee of Musk’s Department of Government Efficiency reportedlystoleAmericans’ personal data from the Social Security Administration and stored it on a thumb drive — the latest accusation ofDOGE-related data leakage.) Chief Pentagon spokesperson Sean Parnell said the department “looks forward to deploying Grok to its official AI platform GenAI.mil in the very near future.” GenAI.milis the military’s secure enterprise platform for generative AI that gives DoD workers access to large language models (LLMs) and other AI tools within government-approved cloud environments. It is designed to help with primarily non-classified tasks like research, document drafting, and data analysis.
View
Jensen Huang just put Nvidia’s Blackwell and Vera Rubin sales projections into the $1 trillion stratosphere
Nvidia CEO Jensen Huang threw out a lot of numbers — mostly of the technical variety — during his keynote Monday to kick off the company’s annualGTC Conferencein San Jose, California. But there was one financial figure that investors surely took notice of: his projection that there will be $1 trillion worth of orders for Nvidia’s Blackwell and Vera Rubin chips, a monetary reflection of a booming AI business. About an hour into his keynote, Huang noted that last year Nvidia saw about $500 billion in demand for its Blackwell and upcoming Rubin chips through 2026. “Now, I don’t know if you guys feel the same way, but $500 billion is an enormous amount of revenue,” he said. “Well, I’m here to tell you that right now where I stand — a few short months after GTC DC, one year after last GTC — right here where I stand, I see through 2027, at least $1 trillion.” The Rubin computing chip architecture, which was first announced in 2024, has been described by Huang as the state of the art in AI hardware that outperforms its Blackwell predecessor. The companysaid in January, when it officially started production of Rubin, it would operate 3.5x faster than the Blackwell architecture on model-training tasks and 5x faster on inference tasks, reaching as high as 50 petaflops. Nvidia has said it expects to ramp up production in the second half of the year.
View
Nvidia’s version of OpenClaw could solve its biggest problem: Security
Nvidia CEO Jensen Huang thinks every company should have anOpenClawstrategy. And Nvidia is here to provide it. Nvidia has developed NemoClaw, an enterprise-grade platform built off the viral, local AI autonomous agent, Huangannouncedduring his GTC keynote on Monday. The open source platform is essentially OpenClaw with enterprise-grade security and privacy considerations baked in. The idea is to turn OpenClaw into a secure platform that enterprises can tap into with one command and control how agents behave and handle data, according to the company. “For the CEOs, the question is, what’s your OpenClaw strategy?” Huang said onstage. “We need it. We all have a Linux strategy. We all needed to have an HTTP HTML strategy, which started the internet. We all needed to have a Kubernetes strategy, which made it possible for mobile cloud to happen. Every company in the world today needs to have an OpenClaw strategy, an agentic systems strategy.” Nvidia worked with OpenClaw’s creator Peter Steinberger to develop NemoClaw, Huang said. Once it is released, NemoClaw users will be able to tap any coding agent or open AI model, including Nvidia’s NemoTron open models to build and deploy AI agents. The platform allows users to access cloud-based models on their local devices. The platform is hardware agnostic — it doesn’t need to run on Nvidia’s own GPUs — and integrates with NeMo, Nvidia’s AI agent software suite. For now, Nvidia is describing NemoClaw as an early-stage Alpha software. “Expect rough edges. We are building toward production-ready sandbox orchestration, but the starting point is getting your own environment up and running,”the company statedon its website in a note directed toward developers. Building enterprise AI agent platforms has become the soup du jour of the AI space in recent months. OpenAI launchedOpenAI Frontier, its open platform for enterprises to build and manage AI agents, in February. In December, global research firm Gartnerreleased a reportabout how governance platforms for AI agents would be the crucial infrastructure needed for enterprises to adopt the AI tech. Nvidia clearly got the message. “OpenClaw gave us, gave the industry exactly what it needed at exactly the time,” Huang said. “Just as Linux gave the industry exactly what it needed at exactly the time, just as Kubernetes showed up at exactly the right time, just as HTML showed up. It made it possible for the entire industry to grab on to this open source stack and go do something with it.”
View
The dictionary sues OpenAI
Encyclopedia Britannica and Merriam-Webster have filed a lawsuit against OpenAI, alleging in itscomplaintthat the AI giant has committed “massive copyright infringement.” Britannica, which owns Merriam-Webster, retains the copyright to nearly 100,000 online articles, which have been scraped and used to train OpenAI’s LLMs without permission, the publisher alleges in the lawsuit. Britannica also accuses OpenAI of violating copyright laws when it generates outputs that contain “full or partial verbatim reproductions” of its content and when the AI lab uses its articles in ChatGPT’s RAG (retrieval augmented generation) workflow. OpenAI’s RAG tool is how the LLM scans the web or other databases for newly updated information when responding to a query. Britannica also alleges that OpenAI violates the Lanham Act, a trademark statute, when it generates made-up hallucinations and attributes them falsely to the publisher. “ChatGPT starves web publishers like [Britannica] of revenue by generating responses to users’ queries that substitute, and directly compete with, the content from publishers like [Britannica],” the lawsuit reads. Britannica also alleges ChatGPT’s hallucinations jeopardize “the public’s continued access to high-quality and trustworthy online information.” Britannica joins a number of other publishers and writers in pursuing legal action against OpenAI over copyright issues. TheNew York Times,Ziff Davis(owner of Mashable, CNET, IGN, PC Mag, and others), and more than a dozennewspapersacross the U.S. andCanada, including the Chicago Tribune, the Denver Post, the Sun Sentinel, the Toronto Star, and the Canadian Broadcasting Corporation, have sued OpenAI. Asimilar Britannica lawsuitagainst Perplexity is still pending. There is not a strong legal precedent that establishes whether using copyrighted content to train an LLM is copyright infringement. But inone particular instance, Anthropic successfully convinced federal judge William Alsup that this use case — using the content as training data — is transformative enough to be legal. However, Alsup argued that Anthropic violated the law by illegally downloading millions of books, rather than paying for them, which warranted a $1.5 billion class action settlement for impacted writers. OpenAI did not respond to TechCrunch’s request for comment before publication.
View
Fuse raises $25M to disrupt aging loan origination systems used by US credit unions
In 2023, after three years of building an automotive lending startup, Fuse co-founders Andres Klaric and Marc Escapa realized that LLMs could modernize something even more significant: the loan origination system (LOS), which is the backbone of the lending industry. Frustrated by the limitations of legacy software, Klaric (pictured left), a Bolivian native, and Escapa (pictured righ), a Spanish immigrant, pivoted their business to build Fuse, an AI-native LOS. On Monday, Fuse announced that it has raised a $25 million Series A led by Footwork, Primary Venture Partners, NextView Ventures, and Commerce Ventures. An LOS serves as the primary system of record for most lenders, managing the entire loan life cycle: from initial application and underwriting to final approval and credit disbursement. However, traditional systems can take as long as a year to integrate and typically have multi-year, expensive contracts, Klaric said. By leveraging AI, Fuse claims its agents can help lenders process higher loan volumes, automate underwriting, and significantly reduce operational costs. The company, which already has over 100 customers, wants to ease credit unions’ transition to Fuse by offering the first 50 qualifying institutions free access to its platform until their current contracts with legacy LOS vendors expire. To support this, the startup has allocated $5 million for a program it’s calling a “rescue fund.” Klaric insists that “it’s not just a marketing gimmick,” explaining that because legacy software costs are high, many credit unions cannot afford to break their current contracts to switch providers. Nikhil Basu Trivedi, a co-founder and general partner at Footwork, told TechCrunch that he backed Fuse because there are over4,000 credit unionsin the United States, and their technology is long overdue for an overhaul. “We know the credit unions are really hurting and want to adopt AI but have no idea how to do it,” he said. Basu Trivedi compared the LOS to an ERP or CRM, noting that it is just as vital to a credit union’s day-to-day operations. He said that swapping out an LOS for another one has traditionally been very difficult. However, as is the case with manyAI ERP-type startups, the founders promise that Fuse can be adopted relatively quickly. Some of the legacy LOS systems that Fuse is trying to displace include publicly traded nCino and private-equity-owned MeridianLink. Naturally, Fuse is not the only startup developing an AI-infused LOS. The company’s competitors include Casca and Glide. Klaric says he strongly believes in the mission of helping credit unions reduce costs in large part because these institutions serve the American middle class. “Credit unions and smaller financial institutions have everything required to win. They have the local presence, the local focus, great member experience. They even have branches in very good locations. The only thing is they don’t they really have is the right technology,” he said.
View
