AI News Archive: June 11, 2026 — Part 5
Sourced from 500+ daily AI sources, scored by relevance.
- Semantic Search for AI Agents at Scale: Retrieval and Ranking for LinkedIn’s Hiring Assistant
Improving LinkedIn's Hiring Assistant with scalable AI-powered search
- How AI is creating an industrial revolution in fraud & deception
AI has changed the speed, cost, scale and variety at which deception and scams can now be produced
Score: 65🌐 MovesJun 11, 2026https://www.rte.ie/brainstorm/2026/0611/1577505-ai-scams-fraud-deception-security-protection-crime/ - Hypha Emerges From Stealth, Announces a $50M Seed Round
Backed by banks, private credit, private equity and real estate insiders, Hypha is turning fragmented investment data into structured intelligence
Score: 65💰 MoneyJun 11, 2026https://www.cityam.com/hypha-emerges-from-stealth-announces-a-50m-seed-round/ - Amazon claims data centers are 7-times more water-efficient than rivals as Seattle pauses new builds
Amazon is working to rewrite the narrative around data center's environmental toll as local communities begin shutting the door on rapid AI expansion. Read More
- AWS launches FinOps agent to bring AI cost governance to cloud spend
AI cost governance is moving to the center of enterprise cloud strategies, and AWS is responding with a new autonomous agent that’s designed to manage the complexity of the change. The FinOps agent, launched in feature preview at FinOps X 2026, monitors cloud costs, detects anomalies, performs root cause analysis and routes alerts directly to […] The post AWS launches FinOps agent to bring AI cost governance to cloud spend appeared first on SiliconANGLE .
Score: 65🌐 MovesJun 11, 2026https://siliconangle.com/2026/06/11/aws-launches-finops-agent-bring-ai-cost-governance-cloud-spend-finopsx/ - GMI Cloud and Magna AI Partner to Expand Global Sovereign AI Infrastructure
GMI Cloud and Magna AI Partner to Expand Global Sovereign AI Infrastructure The Straits Times
- From scams to deepfakes, AI use in Asia creates new cyberthreats
From scams to deepfakes, AI use in Asia creates new cyberthreats Nikkei Asia
- AI sparks alarm in China with call to protect worker rights
AI sparks alarm in China with call to protect worker rights The Japan Times
- CameraMatics secures up to €49 million to scale AI-powered fleet intelligence platform across Europe and the US
CameraMatics, a Dublin-based AI-powered video telematics and fleet intelligence platform, has secured up to €49 million in investment to support its next phase of international expansion and ongoing product innovation as the company scales across the UK, Ireland, Europe, and the US. The investment was secured from a consortium led by Blume Equity, alongside the […] The post CameraMatics secures up to €49 million to scale AI-powered fleet intelligence platform across Europe and the US appeared first on EU-Startups .
- The great AI Irony: China cracks down on Western models while US companies flock to DeepSeek
China is taking a hard-hitting approach to foreign AI and chipmakers, even as Western nations continue to embrace DeepSeek as a cheaper alternative
- AI Summit London: AI’s role in UK defence
The development of artificial intelligence (AI) means that what is new today will inevitably be surpassed in just a few months. Speaking at the AI Summit in London, air chief marshal Rich Knighton, chief of defence staff, told delegates that even today’s AI models have the capacity and capability to transform warfare. “They can process satellite imagery, open source information, logistics, electronic signatures and battlefield reports at a scale that no human headquarters could replicate,” he said. “They could identify patterns, anomalies and even suggest possible courses of action.” Knighton believes that AI models can help commanders understand not only what is happening now, but what might happen next: “I don’t think that we need to fast forward five years or even 35 years from today to see how the battlefield of the future will be shaped by AI.” As Knighton noted, there are now a range of AI systems that are starting to outperform PhD level experts and compete with top-level software engineers. He said the length of time it takes AI to complete tasks autonomously is dramatically reducing: “The frontier is moving incredibly fast and we must be ready to update our assumptions about what AI can do rapidly as it is changing every six months. “We can imagine what this might mean for defence if we can keep pace with the frontier and exploit new models and changes as they are updated every six months or quicker then we will have a clear advantage in the future. Knighton stated that many of today’s AI models already have the potential to accelerate the military decision-making cycle to machine speed, removing what he called “many of the cognitive biases that haunts human decision-making”, adding: “There is both massive risk and huge opportunity even before we think about the ethical questions of the use of AI in warfare.” But he said the UK’s policy remains that humans, not machines, are accountable for decisions, especially when they relate to the application of lethal force. “Defence will continue to ensure that there is a context appropriate human involvement in the development of all AI-enabled systems,” he said. Looking at some of the pilots currently running, Knighton said the Royal Navy has been conducting trials at sea using experimental vessel XV Patrick Blackett. The robotic rigid inflatable boat is equipped with cameras and sensors on board, which feed back data and video to control units and computers on XV Patrick Blackett for analysis. The vessel can be equipped with other sensors and weapons enabling it to be used for intelligence, surveillance and reconnaissance operations with real-time data feeds. “We are using AI to enable fully autonomous navigation and decision-making in un-crewed vessels by fusing sensor data and offering the ability to act without any human input. This is the foundational capability for growing a hybrid navy,” Knighton said. AI is also being used to enhance the effectiveness of military intelligence services to overcome what Knighton describes as “bottlenecked legacy processes and tools”. The result, according to Knighton, is that military analysts have been able to cut identification and response times down from weeks to hours. Read more about AI in warfare In Davos and Washington, AI warfare and AI skills are linked: Lawmakers are discussing AI’s effects on national defense and the economy as well as considering ways to develop a workforce that can meet the challenge. AI chooses nuclear escalation in 95% of simulated crises: With artificial intelligence increasingly deployed in analysis and decision-making in armed conflict, research shows AI systems will not naturally default to ‘safe’ outcomes in nuclear crises.
Score: 65🌐 MovesJun 11, 2026https://www.computerweekly.com/news/366644104/AI-Summit-London-AIs-role-in-UK-defence - Agentic AI Will Industrialize Financial Scams. Are Banks Ready?
Agentic AI Will Industrialize Financial Scams. Are Banks Ready? Boston Consulting Group
Score: 65🌐 MovesJun 11, 2026https://www.bcg.com/publications/2026/how-agentic-ai-will-industrialize-financial-scams - AI's Hidden Energy Bill: Why Visibility is Becoming Critical for Enterprises
Energy consumption is becoming a critical factor in AI investment, governance and sustainability decisions in the U.K.
Score: 65🌐 MovesJun 11, 2026https://aibusiness.com/generative-ai/ai-s-hidden-energy-bill-why-visibility-becoming-critical-enterprises - State-owned AI
Plus sideways CPI
- Jeff Bezos says AI will bring ‘golden ages’ not mass job losses
Amazon founder lays out vision for new $41bn AI lab Prometheus
- Europe at risk of AI-driven irrelevance
Europe’s AI buildout is orders of magnitude too small, a group of AI policy thinkers warned.
Score: 65🌐 MovesJun 11, 2026https://www.semafor.com/article/06/11/2026/europe-at-risk-of-ai-driven-irrelevance - Apple used to partner with ChatGPT. The new Siri shows it's now competing with it.
Apple used to partner with ChatGPT. The new Siri shows it's now competing with it. Business Insider
Score: 65🌐 MovesJun 11, 2026https://www.businessinsider.com/apple-siri-ai-chatgpt-openai-nilay-patel-interview-kafka-2026-6 - 'Oh God, no! Not another thing:' What Anthropic's Mythos-class Fable 5 means for CEOs trying to govern AI
'Oh God, no! Not another thing:' What Anthropic's Mythos-class Fable 5 means for CEOs trying to govern AI Fortune
- Ralliant’s Amir Kazmi On Wiring AI Into Critical Infrastructure
Ralliant's Chief Technology and Growth Officer Amir Kazmi explains how AI-powered workflows, a founder's mindset and a unified role are reshaping precision technology.
- Re-Architecting Die-to-Die IO For AI
Synopsys 3D-IO and the shift to hybrid-bonded 3D integration. The post Re-Architecting Die-to-Die IO For AI appeared first on Semiconductor Engineering .
- Ahead of SpaceX IPO, Elon Musk addresses ASML employees as part of push into chip manufacturing
Elon Musk called ASML a great company in a fireside chat with CEO Christophe Fouquet, as the SpaceX CEO gears up to go big in chip manufacturing.
Score: 65🌐 MovesJun 11, 2026https://www.cnbc.com/2026/06/11/elon-musk-addresses-asml-employees-pushes-into-chip-manufacturing.html - Introducing Act-Two
A new AI model for video generation
- Why global technology firms are betting billions on India's data centres
Rising AI demand, abundant power, favourable policies and strategic geography are turning India into one of the world's fastest-growing data-centre markets
- From Copilots to Clinical Judgment: The Next Phase of AI in Digital Behavioral Health
From Copilots to Clinical Judgment: The Next Phase of AI in Digital Behavioral Health MedCity News
- Databricks Announces OpenSharing, a Protocol for Sharing Data, AI Assets
Databricks has announced the launch of OpenSharing, a new open-source protocol designed to change how organizations share artificial intelligence assets and data. Developed as the next version of the Delta Sharing protocol, OpenSharing is now a project under the Linux Foundation. Delta Sharing, launched by Databricks in 2021, focused on open data sharing across platforms. … continue reading The post Databricks Announces OpenSharing, a Protocol for Sharing Data, AI Assets appeared first on SD Times .
Score: 64🌐 MovesJun 11, 2026https://sdtimes.com/data/databricks-announces-opensharing-a-protocol-for-sharing-data-ai-assets/ - Another sell-off for AI stocks knocks Wall Street back to where it was 5 weeks ago
Another sell-off for AI stocks knocks Wall Street back to where it was 5 weeks ago Dallas News
Score: 64🌐 MovesJun 11, 2026https://www.dallasnews.com/news/world/article/asian-shares-fall-after-a-tech-sell-off-on-wall-22298781.php - Google's DiffusionGemma generates 256 tokens in parallel and self-corrects as it goes
GenAI image generators like Stable Diffusion do not draw a picture pixel by pixel from left to right. They start with noise and iteratively refine the entire image in parallel until it converges, in a process known as diffusion. For years, applying that same principle to text generation had remained out of reach at scale. Standard language models work like a typewriter: one token at a time, left to right, with no ability to revise a committed output. That pattern works in the cloud, where batch sizes keep GPUs saturated. For local inference or low-concurrency deployments, the GPU is idle most of the time. Google's DiffusionGemma, released this week, is an open source experimental model that applies diffusion to text generation at production scale. Built on the Gemma 4 backbone and released under the Apache 2.0 license, it is the first diffusion language model natively supported in the open source vLLM inference platform. It generates a 256-token block in parallel rather than sequentially, with every token position attending to every other. Google says DiffusionGemma generates text up to 4x faster than standard models on GPUs. At batch size 1 on a single Nvidia H100, the FP8 version reaches 1,008 tokens per second. On H200, it hits 1,288 — roughly six times a standard autoregressive baseline, according to vLLM benchmark results published today. Despite the speed gains, Google did not oversell the release. The company's launch post acknowledged directly that DiffusionGemma's overall output quality is lower than standard Gemma 4, adding "For applications that demand maximum quality, we recommend deploying standard Gemma 4." What DiffusionGemma does DiffusionGemma does not generate tokens in order. It starts with a block of 256 random placeholder tokens, effectively a blank canvas, and runs multiple refinement passes over the entire block at once. On each pass, it evaluates every position and locks in the ones it is most confident about. Uncertain positions get randomized and reconsidered on the next pass, with the model using what it resolved in the previous round to inform the next attempt. The block converges progressively until enough positions stabilize to anchor the rest. Two things follow from that architecture. Self-correction. An autoregressive model that commits to a wrong token is stuck with it, because subsequent tokens are already conditioned on the mistake. DiffusionGemma can identify low-confidence positions and re-evaluate them on the next pass. Bidirectional context. Every position attends to every other position in the block simultaneously, including tokens that appear later in the sequence. That makes the model structurally better suited to constrained generation tasks where left-to-right generation fails. Google demonstrated both properties with a fine-tuned Sudoku solver. The base model solved zero puzzles. After fine-tuning on a Sudoku dataset, it reached an 80% success rate and converged in 12 denoising steps rather than 48. The efficiency gain came directly from the model's ability to self-correct and stop early. How it was built DiffusionGemma runs as a 26B Mixture of Experts model that activates only 3.8B parameters during inference. Quantized, it fits within 18GB VRAM on consumer hardware including the Nvidia RTX 4090 and 5090. Google and NVIDIA also optimized for enterprise Hopper and Blackwell servers using NVFP4 kernels. The vLLM integration required new work because DiffusionGemma does not fit the standard serving model. A typical vLLM batch applies the same attention type to every request. DiffusionGemma requests alternate between causal and bidirectional attention as they cycle through prompt reading, canvas refinement and block commit. The team built per-request attention switching into both the Triton and FlashAttention 4 backends and reused the existing speculative decoding path for the refinement loop. The new ModelState interface the team built for this integration is designed to support additional diffusion models in vLLM as they emerge. Where the speed wins and where it does not DiffusionGemma's speed advantage is real but conditional. Where it applies depends entirely on deployment context. The numbers. At batch size 1 on a single H100, vLLM's published benchmarks put the FP8 model at roughly five times a standard autoregressive baseline. On H200, roughly six times. Those peak figures reflect optimal conditions: single user, dedicated hardware, FP8 quantization. Where it wins. Local inference, single-user applications and low-concurrency serving. In those conditions the GPU has spare compute and memory bandwidth is the bottleneck. DiffusionGemma's parallel block generation fills that gap. Where it does not. High-throughput cloud serving. When a server is batching hundreds of concurrent requests, autoregressive models already saturate available compute and DiffusionGemma's parallel decoding provides diminishing returns. The quality ceiling. Guilherme O'Tina, an AI researcher, put a finer point on it on X . "Local artifacts vs hallucinations are different problems and that decides where this actually wins," O'Tina wrote. How it compares Diffusion language models are not new. Researchers have built them at smaller scales for several years, and Inception Labs' Mercury Coder applied the approach commercially to coding tasks in 2025. What DiffusionGemma adds is scale — a 26B MoE backbone, native vLLM serving and a general-purpose instruction-tuned model rather than a domain-specific one. The more useful comparison for engineers evaluating this against existing inference tooling is speculative decoding, and the distinction matters. Speculative decoding keeps a standard autoregressive target model and uses a smaller draft model to guess several tokens ahead. The target model verifies them in one pass. If sampling is correct, the output distribution stays identical to the target. The architecture is unchanged. Andrew Kuncevich , an ML and AI researcher focused on production AI systems, put it directly on X. "DiffusionGemma is different. It does not just guess future tokens. It creates a noisy 256-token canvas and repeatedly denoises the whole block in parallel. So it's not just a decoding trick — it's a different generation paradigm," Kuncevich wrote. Compared to standard Gemma 4, the trade is speed for quality. Google's benchmark data shows DiffusionGemma below standard Gemma 4 on general output quality metrics, with the gap varying by task. On structured constrained tasks, including code infilling, template generation and problems requiring bidirectional constraint propagation, the architecture has a structural advantage that fine-tuning can surface, as the Sudoku result demonstrates. On open-ended generation, standard Gemma 4 remains the stronger option. What this means for enterprises DiffusionGemma serves via a standard vLLM OpenAI-compatible endpoint with no diffusion-specific pipeline changes required. This is not a general-purpose model upgrade. For teams running local or low-concurrency inference, the architecture choice just expanded. Until now, cutting generation latency on dedicated GPU hardware meant using a smaller model and accepting the quality trade-off. DiffusionGemma offers a third path at the same parameter footprint, on consumer hardware, with same-day vLLM support. For constrained generation workloads, bidirectional attention is worth evaluating. Code infilling, structured data generation and tasks where correct output depends on context not yet generated are where this architecture has a structural edge. The ModelState interface built for this integration is designed to generalize as additional diffusion models emerge. The quality trade-off is real and Google acknowledges it. For teams running local inference on dedicated GPU hardware, this is worth testing.
- Anthropic apologizes for invisible Claude Fable guardrails
The company says it will make the covert safeguard preventing model distillation as visible as other safety measures.
- Microsoft's president says Gen Z's AI backlash should be a wake-up call for Big Tech
Microsoft's president says Gen Z's AI backlash should be a wake-up call for Big Tech Business Insider
Score: 63🌐 MovesJun 11, 2026https://www.businessinsider.com/ai-backlash-gen-z-microsoft-president-brad-smith-graduation-speeches-2026-6 - The Indian Workers Training AI Robots To Take Their Jobs
The Indian Workers Training AI Robots To Take Their Jobs Barron's
Score: 63🌐 MovesJun 11, 2026https://www.barrons.com/news/the-indian-workers-training-ai-robots-to-take-their-jobs-814d32b4?refsec - DoorDash lets customers use photos, prompts to order food and book reservations in latest AI push
DoorDash wants to let users simplify the ordering and reservation process through AI.
- White House Honors AI Challenge Winners as Tech Backlash Grows
Students and teachers submitted projects that use AI to solve problems in their schools and communities.
Score: 63🌐 MovesJun 11, 2026https://www.edweek.org/technology/white-house-honors-ai-challenge-winners-as-tech-backlash-grows/2026/06 - Consumers Know, But Don’t Trust Self-Driving Vehicles, Study Shows
A new study reveals consumers are aware of full-automated self-driving vehicle technology, but that's not enough to get them to ride in one.
- MediaTek’s Rally Signals Shift From Laggard to AI Contender
MediaTek Inc. shares are poised for their best quarter on record, as investors bet a shift into artificial intelligence chips can help it shed the overhang of its struggling older-tech business.
Score: 62🌐 MovesJun 11, 2026https://www.bloomberg.com/news/articles/2026-06-11/mediatek-s-rally-signals-shift-from-laggard-to-ai-contender - ‘Do better’: San Francisco is fed up with this Waymo behavior
‘Do better’: San Francisco is fed up with this Waymo behavior San Francisco Chronicle
Score: 62🌐 MovesJun 11, 2026https://www.sfchronicle.com/sf/article/waymos-parking-bike-lanes-22301211.php - DoorDash Launches Conversational AI Assistant
DoorDash Launches Conversational AI Assistant The Information
Score: 62🌐 MovesJun 11, 2026https://www.theinformation.com/briefings/doordash-launches-conversational-ai-assistant - Security Bite: Apple’s most impressive agentic AI feature yet is hiding in the Passwords app
9to5Mac Security Bite is exclusively brought to you by Mosyle, the only Apple Unified Platform . Making Apple devices work-ready and enterprise-safe is all we do. Our unique integrated approach to management and security combines state-of-the-art Apple-specific security solutions for fully automated Hardening & Compliance, Next Generation EDR, AI-powered Zero Trust, and exclusive Privilege Management with the most powerful and modern Apple MDM on the market. The result is a totally automated Apple Unified Platform currently trusted by over 45,000 organizations to make millions of Apple devices work-ready with no effort and at an affordable cost. Request your EXTENDED TRIAL today and understand why Mosyle is everything you need to work with Apple . While WWDC26 is winding down, I’ve had time to reflect on Monday’s keynote, where Apple spent most of its time preaching to parents about on- device Child Safety and, of course, Siri AI . However, it also showcased something insanely neat and ingenious on Apple’s part that is largely being overshadowed. I’m referring to the new agentic AI feature now in iOS 27’s Passwords app .
- Is AI Ready to Replace Your Next Hire — or Will That Backfire?
Is AI Ready to Replace Your Next Hire — or Will That Backfire? entrepreneur.com
Score: 61🌐 MovesJun 11, 2026https://www.entrepreneur.com/leadership/is-ai-ready-to-replace-your-next-hire-or-will-that/503314 - World shares are mixed after another sell-off of AI stocks on Wall St, while oil prices ease
World shares are mixed after another sell-off of AI stocks on Wall St, while oil prices ease Dallas News
Score: 61🌐 MovesJun 11, 2026https://www.dallasnews.com/business/article/asian-shares-slip-after-another-sell-off-of-ai-22300320.php - OpenAI’s Sam Altman to meet Samsung, Kakao, Naver in Korea
OpenAI CEO Sam Altman is returning to South Korea this weekend for talks with Samsung Electronics, Kakao and Naver, as the ChatGPT leader looks to deepen partnerships with Korean technology companies in AI infrastructure, workplace adoption and consumer services. Altman is scheduled to arrive in Korea on Sunday afternoon and begin his main itinerary Monday, according to industry sources on Thursday. His first major stop will be Samsung Electronics’ Digital City campus in Suwon, Gyeonggi Province
- Anthropic announces ‘Claude Corps’ to teach nonprofits to use AI more effectively
Anthropic announces ‘Claude Corps’ to teach nonprofits to use AI more effectively AP News
Score: 61🌐 MovesJun 11, 2026https://apnews.com/article/anthropic-ai-claude-corps-daniela-amodei-b1c130a08417d13e1256f8982d233b0e - Judges are losing patience with lawyers' AI mistakes
Judges are losing patience with lawyers' AI mistakes Business Insider
Score: 61🌐 MovesJun 11, 2026https://www.businessinsider.com/mississippi-judge-removes-lawyers-lawsuit-ai-hallucinations-court-filings-2026-6 - Google DeepMind’s TacticAI can predict football plays 8 seconds before they happen. Palmeiras is the first to use it.
Google DeepMind built an AI that can predict football plays before they happen. TacticAI uses geometric deep learning to model player movement, forecast dynamics up to eight seconds into the future, and recommend tactical adjustments, all from broadcast-style visual data. Brazilian club Palmeiras is the first to use it for live open-play analysis. The system […] This story continues at The Next Web
Score: 60🌐 MovesJun 11, 2026https://thenextweb.com/news/google-deepmind-tacticai-football-palmeiras-predict-plays - SLB strikes deal with Venezuela's PDVSA to modernise oil sector with AI push
SLB strikes deal with Venezuela's PDVSA to modernise oil sector with AI push Reuters
- Microsoft C.E.O. Satya Nadella Says ‘Everyone Is a Stakeholder’ in A.I.
At The New York Times’s Hard Fork Live event, Mr. Nadella addressed the backlash against artificial intelligence and President Trump’s comments about Americans sharing in the wealth of A.I. companies
Score: 60🌐 MovesJun 11, 2026https://www.nytimes.com/2026/06/10/technology/microsoft-satya-nadella-artificial-intelligence.html - Inside Cisco’s Makeover: Targeting AI With Deeper Stack Integration
Cisco has gotten serious about embracing agentic infrastructure operations to provide better ease-of-use and integration of its products.
- The Rise of the 'Botsitters'
The Rise of the 'Botsitters' Business Insider
Score: 60🌐 MovesJun 11, 2026https://www.businessinsider.com/botsitting-ai-hidden-human-labor-at-work-2026-6 - 5 charts from Goldman Sachs show how AI mania compares to 2000 and 2021
5 charts from Goldman Sachs show how AI mania compares to 2000 and 2021 Business Insider
Score: 60🌐 MovesJun 11, 2026https://www.businessinsider.com/ai-bubble-how-it-compares-to-dot-com-2000-crash-2026-6 - Travala Launches World’s First End-to-end Agentic AI Travel Protocol
Travala Launches World’s First End-to-end Agentic AI Travel Protocol markets.businessinsider.com
- AFL-CIO skeptical of government AI stake
“We think that our plan forward is actually a more responsible way to use technology to benefit the people,” AFL-CIO President Liz Shuler said.
Score: 60🌐 MovesJun 11, 2026https://www.semafor.com/article/06/11/2026/afl-cio-skeptical-of-government-ai-stake