Latest AI News
AI Chatbots Tend to Validate Users’ Messages About Suicide and Violence: Study
A new study from researchers at Stanford and other institutions says that artificial intelligence (AI) chatbots often respond to users' messages about suicide and violence by validating their feelings, and in some cases, even encouraging harmful ideas. The research looked at a set of chat logs from people who reported psychological harm linked to chatbot use, and found repeated patterns of chatbots affirming delusional, suicidal, or violent thinking instead of consistently steering users away from it. The study, however, did not name any specific chatbots.
View
Adobe Introduces Custom Models in Firefly, Expands Access to Project Moonlight
Adobe, on Thursday, introduced several new offerings for its Firefly artificial intelligence (AI) Studio. The biggest highlight from the San Jose, California-based software giant is Custom Models, a new way to let users train the AI using their images for style and character consistency. Additionally, the company also introduced a new video tool and expanded access to AI models via its platform. Adobe's Project Moonlight, which was first announced in October 2025, is now also being expanded to more users in private beta.
View
Amazon brings Alexa+ to the UK
Amazon is bringing its new AI-powered conversational assistant, Alexa+, to the U.K., the first country to get the AI assistant outside of North America. The company is currently letting users in the U.K. try out Alexa+ for free via an early access program. Users whobuy the new Amazon Echowill receive an invite for the early access program, and the company says it plans to enable Alexa+ for “hundreds of thousands” of customers in the coming weeks. Once the early access program ends, Prime subscribers will get to use Alexa+ for free, and non-Prime users will need to pay £19.99 per month. The company hasn’t specified when the early access program ends. Alexa+ works with Amazon’s new Echo devices, Fire TV, and the Alexa app, and can carry conversational context from one device to another. The company plans to extend support for Alexa+ to browsers as well. Amazon said it has customized Alexa+ for U.K. customers, and the assistant can understand local context as well as commonly used phrases. “Local teams, including engineers, linguists, and speech scientists at Amazon’s Tech Hub in Cambridge, UK, have used different techniques—such as reinforcement learning, accent-neutral speech representations, and regional embeddings—to make sure Alexa+ genuinely understands British customers,” the companysaid in a blog post. Amazon noted that users can ask for suggestions from services such as OpenTable, JustEat, and Treatwell. They can also get their news from sources like The Independent, The Guardian, Press Association, and Future Publishing. Amazon first unveiledAlexa+ in February 2025, but its rollout has been slow. By June, the company had just givenone million people access to the new assistant in the U.S.Last month, it finally madeAlexa+ available to all U.S.-based users. The company had also launched early access programs in Canada and Mexico. Last month, the company introduced new“personality” options for Alexa+to let users customize the tone of responses. Earlier this month,Amazon rolled out an adult-only “Sassy” mode, but said that the assistant won’t support NSFW content.
View
TechCrunch Startup Battlefield 200 nominations are still open
Pre-Series A founders, this is your reminder! Nominations forStartup Battlefield 200are still open, and the arena is already filling up. If you’ve been planning to apply, now’s the time to move. AtTechCrunch Disrupt 2026, hand-selected startups won’t just pitch. They’ll step onto the stage and go head-to-head in front of world-class VCs and the full TechCrunch audience. This is where startups are tested, challenged, and pushed into the spotlight. Nominate your startup or put forward one that’s ready to step into the arena and prove it belongs.Submit here. Global exposure and $100,000 in equity-free funding are on the line. So is direct access to leading VCs. And the chance to prove your startup belongs among the best. Companies like Trello, Mint, Dropbox, Discord, and Fitbit once stood exactly where you are now. They entered the startup arena and made their mark. Nominations close May 27, but the strongest founders don’t wait for the clock. They move early, prepare harder, and take their shot before the crowd catches up. Enter Startup Battlefield 200 now.Start nominating.
View
Accenture’s $22 Billion Bookings in Q2 Reflect an AI Demand, Not AI Boom
Accenture saw an increase in large deals, but AI is not yet driving a broad-based spending cycle.
View
Multiverse Computing pushes its compressed AI models into the mainstream
With private company defaults running atupwards of 9.2%— the highest rate in years — VC firm Lux Capital recently advised companies relying on AI to get their compute capacity commitmentsconfirmed in writing. With financial instability rippling through the AI supply chain, Lux warned, a handshake agreement isn’t enough. But there’s another option entirely, which is to stop relying on external compute infrastructure altogether. Smaller AI models that run directly on a user’s own device — no data center, no cloud provider, no counterparty risk — are getting good enough to be worth considering. AndMultiverse Computingis raising its hand. The Spanish startup has so far kept a lower profile than some of its peers, but as demand for AI efficiency grows, this is changing. After compressing models from major AI labs including OpenAI, Meta, DeepSeek and Mistral AI, it has launched both an app that showcases the capabilities of its compressed models and an API portal — a gateway that lets developers access and build with those models — that makes them more widely available. TheCompactifAI app, which shares its name with Multiverse’s quantum-inspired compression technology, is an AI chat tool in the vein of ChatGPT or Mistral’s Le Chat. Ask a question, and the model answers. The difference is that Multiverse embedded Gilda, a model so small that it can run locally and offline, according to the company. For end users, this is a taste of AI on the edge, with data that doesn’t leave their devices and doesn’t require a connection. But there’s a caveat: their mobile devices must have enough RAM and storage. If they don’t — and many older iPhones won’t — the app switches back to cloud-based models via API. The routing between local and cloud processing is handled automatically by a system Multiverse has named Ash Nazg, whose name will ring a bell for Tolkien fans as it references the One Ring inscription in “The Lord of the Rings.” But when the app routes to the cloud, it loses its main privacy edge in the process. These limitations mean that CompactifAI is not quite ready for mass customer adoption yet, although that may never have been the goal. According to data from Sensor Tower, the app hadfewer than 5,000 downloadsin the past month. The real target is businesses. Today, Multiverse is launching aself-serve API portalthat gives developers and enterprises direct access to its compressed models — no AWS Marketplace required. “The CompactifAI API portal [now] gives developers direct access to compressed models with the transparency and control needed to run them in production,” CEO Enrique Lizaso said in a statement. Real-time usage monitoring is one of the key features of the API, and that’s no accident. Alongside the potential advantages of deploying on the edge, lower compute costs are one of the main reasons why enterprises are considering smaller models as an alternative to large language models (LLMs). It also helps that small models are less limited than they used to be. Earlier this week, Mistral updated its small model family with thelaunch of Mistral Small 4, which it says is simultaneously optimized for general chat, coding, agentic tasks and reasoning. The French company alsoreleased Forge, a system that lets enterprises build custom models, including small models for which they can pick the tradeoffs their use cases can best tolerate. Multiverse’s recent results also suggest the gap with LLMs is narrowing. Its latest compressed model,HyperNova 60B 2602, is built on gpt-oss-120b — an OpenAI model whose underlying code is publicly available. The company claims it now deliversfaster responsesat lower cost than the original it was derived from, an advantage that matters particularly for agentic coding workflows, where AI autonomously completes complex, multi-step programming tasks. Making models small enough to operate on mobile devices while still remaining useful is a big challenge.Apple Intelligencesidestepped that issue by combining an on-device model and a cloud model. Multiverse’s CompactifAI app can also route requests to gpt-oss-120b via API, but its main goal is to showcase that local models like Gilda and its future replacements have advantages that go beyond cost savings. For workers in critical fields, a model that can run locally and without connecting to the cloud offers more privacy and resilience. But the bigger value is in the business use cases this can unlock – for instance, embedding AI in drones, satellites, and other settings where connectivity can’t be taken for granted. The company already serves more than 100 global customers including the Bank of Canada, Bosch and Iberdrola, but expanding its customer base could help it unlock more funding. After raising a$215 million Series Blast year, it is nowrumored to be raising a fresh €500 million funding roundat a valuation of more than €1.5 billion.
View
How India Became Volkswagen's Command Centre for Software-Defined Vehicles
Embitel, acquired by Volkswagen in 2022, operates on a “GCC Plus” model that blends captive centre capabilities with external market exposure.
View
OpenAI Unveils 16MB, 10-Min Model Training Challenge on NVIDIA H100 GPUs
The Model challenge is designed to test the ability to tackle unfamiliar problems with creativity and rigour, OpenAI said.
View
Siemens Says Agentic AI Is Early, Doubles Down on Patents
‘Agentic AI is still in its infancy,’ says CTO Peter Koerte, as Siemens expands internal AI platforms and builds a portfolio of over 1,000 AI and data science patents.
View
This Sports Tech Startup Feeds 'Post-Match Analysis' Into Live Games
SportVot uses computer vision, cloud infra, and edge computing to make live sports broadcasting and analytics accessible to grassroots tournaments.
View
Can AI End Monsoon Guesswork on Indian Farms?
Language is the primary barrier to significant adoption of AI at scale.
View
Microsoft Pauses Automatic Rollout of Microsoft 365 Copilot App on Windows
Microsoft has temporarily stopped the automatic rollout of the Microsoft 365 Copilot app on Windows devices with Microsoft 365 desktop apps. The move comes as the company adjusts its approach to integrating AI features more broadly across Windows and its productivity ecosystem. While Copilot has been a key part of Microsoft's recent software strategy, the latest update suggests a more cautious rollout. The company has not confirmed when the automatic installation process will resume.
View
