AI News Archive: June 25, 2026 — Part 3
Sourced from 500+ daily AI sources, scored by relevance.
- TrueFoundry acquires MLOps pioneer Seldon AI to accelerate enterprise agentic AI
TrueFoundry Inc., a startup providing management for artificial intelligence workloads, announced Wednesday that it has acquired open-source, cloud-agnostic enterprise machine learning platform Seldon Technologies Ltd. for an undisclosed price. The company provides the infrastructure needed to run AI models in production without the pain of building it in-house by delivering prepackaged features for deploying and […] The post TrueFoundry acquires MLOps pioneer Seldon AI to accelerate enterprise agentic AI appeared first on SiliconANGLE .
- Micron Is the New Nvidia, It Just Changed the AI Game
Micron Is the New Nvidia, It Just Changed the AI Game Barron's
Score: 70🌐 MovesJun 25, 2026https://www.barrons.com/articles/micron-nvidia-ai-stock-market-things-to-know-today-9364157a - Ferroelectric memory enables one chip to sample randomness and compute for generative AI
For the first time, a research team has demonstrated an artificial intelligence semiconductor technology that integrates the core functions of generative AI into a single device platform based on ferroelectric memory. This technology is significant as the first demonstration of implementing the two essential functions required for generative AI—random sampling and stable computation—within a single memory array.
Score: 70🌐 MovesJun 25, 2026https://techxplore.com/news/2026-06-ferroelectric-memory-enables-chip-sample.html - Adobe snags Topaz Labs as it expands AI capabilities across Creative Cloud
The acquisition expands Adobe's Firefly generative AI tools and brings expertise in AI models that run on local devices rather than relying on cloud computing.
Score: 70🌐 MovesJun 25, 2026https://www.bizjournals.com/sanjose/news/2026/06/25/adobe-topaz-labs-acquisition.html?ana=brss_6150 - OfficeSpace Reports an Estimated 600%+ Growth in Demand for its AI Capabilities Since January as Enterprises Turn to Agentic Systems to Manage the Built World
OfficeSpace Reports an Estimated 600%+ Growth in Demand for its AI Capabilities Since January as Enterprises Turn to Agentic Systems to Manage the Built World Toronto Star
- Scaling world understanding for autonomous systems without equivalent cost scaling
Scaling world understanding for autonomous systems without equivalent cost scaling General Motors
Score: 70🌐 MovesJun 25, 2026https://news.gm.com/home.detail.html/Pages/news/us/en/engineering/2026/jun/0625-scaling-autonomous-systems.html - Water joins energy as top AI flashpoint
Water is fast becoming one of the defining fights around the AI buildout. Why it matters: After spending much of the past year defending data centers' electricity demands, major tech companies driving the AI boom are increasingly making the case that their water use is manageable too. Driving the news: Over the past several weeks, Google , Amazon and Microsoft have each launched new efforts to explain and justify the water use of their AI infrastructure, highlighting measures such as water replenishment projects, recycled-water use and new cooling technologies. Nvidia — the world's dominant AI chip maker — claimed this week that water concerns could be largely addressed by its latest generation of technology. What they're saying: "The growing conversation about water and energy use by data centers has forced these companies to scramble, to rethink what they're doing and to become more transparent about what they're doing," said Peter Gleick, co-founder of the Pacific Institute, a California-based water research nonprofit, and one of the nation's leading water experts. "They're starting to understand the reputational risk of the massive rollout of data centers that have big energy and water footprints." Friction point: Roughly 70% of people in the U.S. said they would oppose data centers in their communities, with equal weight placed on water and energy use as top concerns, according to Gallup polling from May . State of play: Such worries are infiltrating debates at all levels around the world. The United Nations Secretary General António Guterres called for more transparency on data centers' energy, water and land use in a speech earlier this week in London. Also this week, lawmakers in Virginia — which has the world's highest number of data centers — moved toward clamping down on the most water-intensive methods of cooling. Reality check: Compared to other major industries, data centers actually use far less water — a point tech executives are quick to point out and some independent experts agree with. Data: Cleanview analysis of government, industry and academic sources, including 2024 Lawrence Berkeley National Laboratory study ; Note: Power plants include fossil fuel and nuclear facilities. Data centers include on-site cooling and associated electricity generation; Chart: Amy Harder/Axios "The projections for water demand are not eyebrow-raising," said Sarah Porter, director of the Kyl Center for Water Policy at Arizona State University. Concerns about water are largely a "substitute for concerns people have for this fast-developing industry." Yes, but: Experts, including both Gleick and Porter, caution that aggregate water-use figures can obscure local impacts, particularly in drought-prone regions where even modest demand can become contentious. "The important point is: How much water does a data center use in the region where it's taking the water from?" Gleick said. Comparisons to other industries also may do little to ease concerns in communities facing the prospect of a big new industrial neighbor. How it works : Energy and water are intricately — and sometimes inversely — connected. Water-based cooling systems generally use less electricity than air-based systems, creating a tradeoff between water consumption and energy demand. Generating the electricity that powers data centers requires water as well if it's powered by fossil fuels or nuclear power. Wind and solar require no water. Zoom in: Water-intensive cooling has historically been favored because it uses less energy and is less expensive, but it is facing growing public opposition. "However, the court of public opinion has spoken loudly that consuming water for cooling on data centers is no longer an acceptable method," said Aaron Bilyeu, chief development officer of Cloverleaf Infrastructure, a data center developer. Zoom out: For all the focus on cooling technology, much of a data center's broader water footprint comes from the electricity it consumes rather than water used directly at the facility. A recent Bank of America report estimated electricity generation accounts for roughly 75% of a data center's total water footprint. What's next: Transparency is emerging as a key next phase of AI water worries. Tech giants, including Microsoft and Google, are set to release annual environmental reports in the coming weeks that could shed more light on their water use. What we're watching: Guterres added urgency to those moves when he proposed an AI environmental transparency initiative this week. "I am calling on every major AI company to measure and publicly disclose the full environmental impact of its systems — carbon, water and land footprints," Guterres said.
- US is ‘superhero’, China ‘supervillain’ in global AI contest, American officials warn
US House Foreign Affairs Committee Chairman Brian Mast warned that “America is the superhero” and China the “supervillain” in the contest for global artificial intelligence (AI) leadership on Thursday, just two days after US Treasury Secretary Scott Bessent said America’s “biggest risk” on AI is China getting ahead. The United States and China remain locked in an increasingly competitive race for worldwide AI supremacy, with many American officials concerned that China is eroding the US’ early...
- The New Push to Ready Millions for AI Career Upheaval
A coalition of employers and state governments says it is developing a sweeping strategy to help workers respond to the AI age.
- Small Language Models Outperform Frontier AI On Cost, Speed And Accuracy
Bigger has defined AI from day one. New data says task-specific small models beat frontier LLMs on accuracy, cost and speed — and save money.
- Micron overtakes Meta, Tesla in market value amid relentless AI infrastructure demand
Micron overtakes Meta, Tesla in market value amid relentless AI infrastructure demand Reuters
- SpaceX’s $2.5 Trillion Value Is Built on AI Data Centers. Their Cost Is Otherworldly.
SpaceX’s $2.5 Trillion Value Is Built on AI Data Centers. Their Cost Is Otherworldly. Barron's
Score: 69🌐 MovesJun 25, 2026https://www.barrons.com/articles/spacex-orbital-data-centers-elon-musk-d0c67f58 - Acer Introduces Aspire X 16 AI, Aspire 18 AI and New Copilot+ Desktops
Acer Introduces Aspire X 16 AI, Aspire 18 AI and New Copilot+ Desktops PCMag Middle East
Score: 69🌐 MovesJun 25, 2026https://me.pcmag.com/en/laptops/37545/acer-introduces-aspire-x-16-ai-aspire-18-ai-and-new-copilot-desktops - From solar power to AI: How Dubai is future-proofing vital services
From solar power to AI: How Dubai is future-proofing vital services Gulf News
Score: 69🌐 MovesJun 25, 2026https://gulfnews.com/uae/from-solar-power-to-ai-how-dubai-is-future-proofing-vital-services-1.500586359 - Summer Davos: Leaders on how to make the AI payoff happen for everyone
Throughout the Annual Meeting of the New Champions 2026, leaders considered the challenges of AI transformation. From deploying infrastructure to data readiness, here are four key takeaways.
Score: 69🌐 MovesJun 25, 2026https://www.weforum.org/stories/2026/06/how-to-make-the-ai-payoff-happen-leaders-at-summer-davos/ - Most major AI chatbots still lean left on political questions, even "anti-woke" models are no exception
A Washington Post investigation shows that most major AI chatbots still skew left on political questions. OpenAI's GPT-5.5 gave exclusively left-leaning arguments 80 percent of the time, and even Musk's Grok, marketed as anti-"woke," leaned left more often than not. The one outlier: Google's Gemini 3.1 Pro presented both sides 93 percent of the time. The article Most major AI chatbots still lean left on political questions, even "anti-woke" models are no exception appeared first on The Decoder .
- India’s military speeds up orders for weaponised drones
Conflict in Iran pushes New Delhi to focus on domestic defence supply chains
- ADNOC Drilling delivers first AI walking rig early
ADNOC Drilling delivers first AI walking rig early Gulf News
Score: 69🌐 MovesJun 25, 2026https://gulfnews.com/business/energy/adnoc-drilling-delivers-first-ai-walking-rig-early-1.500586159 - Nvidia co-founder to keynote Korea-U.S. science conference
Nvidia co-founder to keynote Korea-U.S. science conference 매일경제
- Generative AI designs DNA origami to match user-drawn shapes automatically
A joint research team has developed an automated design technology that enables the creation of DNA origami structures that exactly match user-drawn shapes using generative AI. The generative design model, "Generative SNUPI," arranges DNA bases along the contour of a user-defined shape and automatically designs the bonding pathways required to assemble the structure. This breakthrough effectively enables AI to function as a nanodesigner.
- Baidu Releases Unlimited OCR, a 3B Model That Keeps the KV Cache Flat for Long-Document Parsing
Baidu Releases Unlimited OCR, a 3B Model That Keeps the KV Cache Flat for Long-Document Parsing MarkTechPost
- DeepReinforce Releases Ornith-1.0: An Open-Source Coding Model Family That Learns Its Own RL Scaffolds
DeepReinforce Releases Ornith-1.0: An Open-Source Coding Model Family That Learns Its Own RL Scaffolds MarkTechPost
- AI has helped to slash nuclear licensing review times, NRC official says
Some reviews that once took four years to complete are done in nine months, NRC Chief Data Officer and Deputy Chief AI Officer Basia Sall said on Thursday.
- Liquid AI's smallest model yet LFM2.5-230M beats models 4X its size at data extraction, can run 'anywhere'
Liquid AI, founded by former MIT computer scientists, today released its smallest AI language model yet, LFM2.5-230M , and enterprises would do well to consider it for their uses in data extraction and local deployment on smartphones, laptops and robotics. This is a 230-million-parameter foundation model explicitly designed for on-device agentic workflows, and as Liquid states in its release blog post, that small size makes it possible to run nearly "anywhere." According to Liquid, it also outperforms models more than 4X its size on selected benchmarks, specifically doing better at data extraction than the 800 million parameter count Alibaba Qwen3.5-0.8B (Instruct) and 1-billion parameter Google Gemma 3 1B. The model targets developers and engineers building lightweight data extraction pipelines and autonomous edge systems. Operating under a dual-use commercial license, the model remains free for individuals and companies generating less than $10 million in annual revenue, while requiring a paid enterprise agreement for larger corporations. This release distinguishes itself from other small AI models by utilizing the LFM2 architecture to achieve high inference speeds without the massive memory overhead typical of parameter-heavy transformers. While major AI companies Anthropic, OpenAI, Google, Microsoft, Meta and others push parameter counts into the hundreds of billions or trillions to achieve frontier performance, a parallel race focuses entirely on the edge and local deployments. Liquid AI's launch of LFM2.5-230M signals a pivotal shift toward architectural efficiency over brute-force scaling. By squeezing 19 trillion tokens of pre-training into a 230-million-parameter footprint, the company demonstrates that edge devices do not need massive computational power or persistent cloud connections to execute complex, multi-step agentic workflows. How LFM2.5-230M works The LFM2.5-230M model diverges from standard transformer architectures, relying instead on the LFM2 framework. This architecture functions as a hybrid system, interleaving gated short-range convolutions with grouped-query attention to process information efficiently. For those tracking the evolution of efficient architectures, Liquid’s approach shares a similar conceptual goal: managing long contexts and sequential data effectively on edge hardware without the quadratic memory costs of pure attention mechanisms. The model supports an expansive 32K context window, allowing it to ingest substantial documents or continuous streams of robotic telemetry. When analyzing the performance charts provided in the release, the architectural efficiency becomes visually apparent. The model maintains a memory footprint of under 400MB while achieving prefill and decode speeds that outpace comparable models like Gemma 3 1B IT and Granite 4.0-H-350M. On a Samsung Galaxy S25 Ultra equipped with a Qualcomm Snapdragon Gen4 CPU, the model reaches a decode speed of 213 tokens per second. Even on a highly constrained Raspberry Pi 5, the model maintains a decode rate of 42 tokens per second. Furthermore, internal benchmarking shows the GPU inference stack delivers lower end-to-end latency than competing small models across all concurrency levels. Why it matters for enterprises To understand why a 230-million-parameter model is necessary, one must look at how enterprises currently manage data. Organizations have traditionally relied on rigid, rule-based Extract, Transform, Load (ETL) scripts to move and process data. However, these legacy systems are notoriously brittle; a simple change in a document's layout or a schema update can break the entire pipeline. To solve this, the industry is shifting toward "AI ETL," where machine learning infers mappings, detects schema drift, and adapts to changes automatically. In a modern lightweight data extraction pipeline, an AI model connects to unstructured sources—like PDFs, emails, or web forms—and structures the data into formats like JSON without requiring hardcoded rules. For enterprises, using a massive flagship model like Claude Opus 4.6 (which costs $5.00 per million input tokens) to parse routine invoices, format addresses, or route telemetry data is economically unviable. This is where models like LFM2.5-230M become critical. Designed explicitly as a lightweight extraction engine, it allows companies to automate repetitive formatting and data parsing at a fraction of the compute cost and latency, running directly on local hardware rather than relying on expensive, continuous cloud API calls. Small Model Benchmarks: LFM vs. The 3B Class The AI industry in mid-2026 is seeing a renaissance in "small" models, but the definition of "small" varies wildly. Recently, the open-weight community was stunned by Weibo's VibeThinker-3B, a 3-billion-parameter model built on a Qwen2-style backbone that achieved a massive 94.3 on the AIME 2026 math benchmark, rivaling 600-billion-parameter behemoths through aggressive data curation and reinforcement learning. Similarly, Google's Gemma 4 family — which recently crossed 200 million downloads — pushes frontier AI to the edge, including the E2B (2 billion parameters) designed specifically for mobile and IoT deployments. By contrast, Liquid AI's LFM2.5-230M operates in a completely different weight class. At just 230 million parameters, it is roughly one-tenth the size of Google's smallest Gemma 4 model and VibeThinker-3B. Because of its microscopic footprint, LFM2.5-230M is not designed to compete on reasoning-heavy workloads like advanced math, coding, or creative writing—a constraint Liquid AI explicitly acknowledges. However, in its intended domains of data extraction and tool calling, the model punches well above its weight class. Benchmarks released by Liquid AI show LFM2.5-230M scoring 43.26 on the BFCLv3 tool-use benchmark, dominating IBM's Granite 4.0-350M (39.58) and completely outpacing larger 1-billion-parameter models like Google's Gemma 3 1B IT (16.61). On CaseReportBench for data extraction, it scores 22.51, decimating the Qwen3.5-0.8B (Instruct). LFM2.5-230M proves that while 3-billion-parameter models like VibeThinker are solving advanced calculus, a 230-million-parameter model is the superior, highly optimized choice for executing structured tool calls and keeping agentic pipelines running efficiently on constrained hardware. Advanced research uses Because it excels at tool calling, LFM2.5-230M functions primarily as a skill-selection layer. Liquid AI demonstrated this capability by deploying the model on a Unitree G1 humanoid robot. Running entirely on-device via the robot's onboard NVIDIA Jetson Orin compute module, the model successfully processes complex environmental commands. As noted in the company's technical blog, the model takes a free-form instruction like, *"Hold still for 2 seconds, then walk forward at 1 meter per second for 3 meters, hold a forward one-leg kneel for 5 seconds, and walk backward at 0.5 meters per second for 3 meters,"* and automatically translates it into a structured multi-step plan calling on pre-trained low-level skills provided by NVIDIA's SONIC framework. The base and post-trained models are available immediately on Hugging Face, with native day-one support across the inference ecosystem for llama.cpp (GGUF), MLX, vLLM, SGLang, and ONNX. Dual-use, custom LFM Open License Liquid AI ships LFM2.5-230M under the LFM Open License v1.0. Despite the word "open" in the title, this is not an Open Source Initiative (OSI) compliant license; it operates as a restricted, dual-use commercial framework. For independent developers, researchers, and early-stage startups, the license functions identically to open-source software. Users receive a perpetual, worldwide, royalty-free license to reproduce, modify, and distribute the model, provided they retain original copyright notices and prominently state any modifications. However, the license includes a strict "Commercial Use Limitation". Any legal entity generating $10 million or more in annual revenue loses the right to use the model commercially under this agreement. Large enterprises crossing this financial threshold must negotiate a separate, paid commercial agreement with Liquid AI to deploy the model in production. This strategy protects the company from having its intellectual property absorbed by major technology conglomerates for free, while still seeding the model at the grassroots developer level.
- BlackLine® Expands Agentic Financial Operations Platform; Establishing the Trust Infrastructure for AI-Powered Finance
BlackLine® Expands Agentic Financial Operations Platform; Establishing the Trust Infrastructure for AI-Powered Finance Toronto Star
- Swarm robots inspired by bees and ants could transform the future of mining
Researchers at Adelaide University have developed a new type of robotic system inspired by bees and ants that could make mining safer, more efficient and more sustainable.
- Robot dentist prepares tooth for a crown
The tiny robo-dentist will see you now. The post Robot dentist prepares tooth for a crown appeared first on Popular Science .
- Oracle partners with Theator to expand AI-powered surgical documentation to operating rooms
Oracle Health, owned by tech giant Oracle after it acquired EHR company Cerner in 2022, is ramping up its AI capabilities while also expanding its AI ecosystem through partnerships.
- HCLTech and ServiceNow join forces to scale enterprise AI with Google Cloud
HCLTech has expanded its collaboration with Google Cloud and ServiceNow to deliver AI agents for enterprise adoption on the Gemini Enterprise platform. The launch coincides with HCLTech’s sponsorship of the Sydney Google Cloud Summit 2026. Building on HCLTech’s recently launched Gemini Enterprise business unit, the latest collaboration brings together advanced ServiceNow AI capabilities, enterprise workflow orchestration and […] The post HCLTech and ServiceNow join forces to scale enterprise AI with Google Cloud appeared first on CXOToday.com .
- Reward hacking is swamping model intelligence gains
Explores how reward hacking undermines gains in model intelligence.
- Andhra Pradesh, Airbound target one of the world’s largest drone delivery networks
Andhra Pradesh, Airbound target one of the world’s largest drone delivery networks YourStory.com
Score: 68🌐 MovesJun 25, 2026https://yourstory.com/2026/06/andhra-pradesh-airbound-target-large-drone-delivery-network - AI Sales Start to Justify Data-Center Spending Boom, Report Says
Revenue from AI has reached a tipping point, showing that the hundreds of billions of dollars tech companies are spending on it may be economically sustainable, according to a report from research firm Exponential View. Azeem Azhar, founder of Exponential View, joins Ed Ludlow on "Bloomberg Tech" to discuss the findings of this report. (Source: Bloomberg)
Score: 68🌐 MovesJun 25, 2026https://www.bloomberg.com/news/videos/2026-06-25/ai-sales-start-to-justify-data-center-spending-boom-video - Salesforce Launches Agentforce Help Agent That Deploys in Minutes and Only Charges for Resolutions
Less than two years ago, Salesforce launched Agentforce, the platform to build, deploy, and govern AI agents. Companies built service agents on it, and many work beautifully. But building one was real work: you connected your own knowledge, defined your own actions, and wired up each channel yourself. Today, that work is done for you. […]
Score: 68🌐 MovesJun 25, 2026https://www.salesforce.com/news/stories/agentforce-help-agent-announcement/ - Anthropic’s Claude is winning over paid consumers, a market owned by ChatGPT
Despite ChatGPT's commanding market lead, consumers who pay for AI have been increasingly choosing Anthropic's Claude, data shows.
Score: 68🌐 MovesJun 25, 2026https://techcrunch.com/2026/06/25/anthropics-claude-is-winning-over-paid-consumers-a-market-owned-by-chatgpt/ - Westwell Demonstrates How AI-Powered Autonomous Freight Is Transforming Border Logistics
Westwell Demonstrates How AI-Powered Autonomous Freight Is Transforming Border Logistics The Straits Times
- Can LLM Agents Select and Engage with Biological Tools?
Large language model agents are capable of performing initial interactions with biological tools (BTs), indicating that agents could lower expertise barriers for actors seeking to use BTs. Targeted testing of agent design capabilities is warranted.
- Apple just told consumers they need to foot the bill for AI data centers
Apple just told consumers they need to foot the bill for AI data centers Business Insider
Score: 67🌐 MovesJun 25, 2026https://www.businessinsider.com/apple-ai-price-hikes-data-centers-what-it-means-2026-6 - The Gulf Between AI Progress and Political Understanding (with Dex Hunter-Torricke)
The Gulf Between AI Progress and Political Understanding (with Dex Hunter-Torricke)
- Hacking, AI & Terrorism: A Lethal Triad?
Hacking, AI & Terrorism: A Lethal Triad? Oxford Lifelong Learning
Score: 67🌐 MovesJun 25, 2026https://lifelong-learning.ox.ac.uk/courses/hacking-ai-terrorism-a-lethal-triad/ - ByteDance's Doubao Crosses Production-Grade Threshold with 180 Trillion Daily Tokens
On June 23, ByteDance officially launched Doubao-Seed-2.1 Pro (Doubao 2.1 Pro), its flagship large language model that has pushed daily token calls to a stagger...
- Global chip stocks jump as blowout Micron results reignite AI rally
Global chip stocks jump as blowout Micron results reignite AI rally Reuters
- Paystack’s new experiment lets AI agents make everyday payments for users
Paystack has launched Paystack Index, an experimental checkout product that lets users complete everyday transactions through AI agents, marking an early bet on agentic commerce in Africa.
- Agentic AI bot helps scientists speak to robots, speeding up experiments
Researchers at the Department of Energy's Pacific Northwest National Laboratory use a slew of autonomous robots to design and implement experiments. However, setting up an experiment on an autonomous lab robot is surprisingly slow. The effort requires a lengthy back-and-forth between a scientist and an engineer to design the experimental steps—a process that can take weeks.
Score: 66🌐 MovesJun 25, 2026https://techxplore.com/news/2026-06-agentic-ai-bot-scientists-robots.html - X user tricks Grok into sending them $200,000 in crypto using morse code
An X user managed to trick AI chatbot Grok into sending around $200,000 worth of crypto after exploiting its link with an automated trading bot. The incident involved Grok and 'Bankrbot', two AI systems with wallet access, which were manip ... (https://incidentdatabase.ai/cite/1556#7454)
- Qualcomm forecasts $15 billion in data centre chip sales by 2029
UPDATE 2-Qualcomm forecasts $15 billion data center chip sales by 2029, shares soar
Score: 66🌐 MovesJun 25, 2026https://www.khaleejtimes.com/business/tech/qualcomm-forecasts-15-billion-in-data-centre-chip-sales-by-2029 - News, cultural groups want clarity on copyright after Ottawa releases AI strategy
News, cultural groups want clarity on copyright after Ottawa releases AI strategy Toronto Star
- Anthropic hires Orange's AI chief amid Europe push
Anthropic hires Orange's AI chief amid Europe push Reuters
Score: 66🌐 MovesJun 25, 2026https://www.reuters.com/business/media-telecom/anthropic-hires-oranges-ai-chief-amid-europe-push-2026-06-25/ - After Anthropic shutdown, China's Z.ai closes frontier gap as it plans dual listing
After Anthropic shutdown, China's Z.ai closes frontier gap as it plans dual listing Reuters
- Tech Brief (June 25): ByteDance’s AI Assistant Doubao Launches Paid Version
Tech Brief (June 25): ByteDance’s AI Assistant Doubao Launches Paid Version Caixin Global
- Chinese AI firms scale aggressively to compete with US
DeepSeek hopes to at least double the size of all its departments to become a “driving force” in developing artificial general intelligence.
Score: 65🌐 MovesJun 25, 2026https://www.semafor.com/article/06/25/2026/chinese-ai-firms-scale-aggressively-to-compete-with-us