OpenAI for Government
A new program bringing ChatGPT, DALL·E, and API tools directly to U.S. federal, state, and local agencies. Kicks off with a $200 M DoD pilot, cutting bureaucracy and saving Pennsylvania staff ~2 hrs/day on routine tasks.
Source

Chatbot-Driven Disinformation in LA Protests
WIRED’s Uncanny Valley podcast uncovers how malicious bots stoked violence and confusion in the latest Los Angeles demonstrations.
Source

Yupp’s Feedback-Paying Chatbot Tool
On the Yupp platform, users earn $50/mo to rate AI prompts—data now fueling model fine-tuning across industry labs.
Source

IBM’s Quantum Roadmap
Announced this week: IBM aims for a large-scale, error-corrected quantum computer by 2028, promising to offload classical‐compute bottlenecks for future AI workloads.
Source

Manus Ignites China’s AI Agent Boom
After its release, countless Chinese startups cloned the general-purpose AI agent concept, leveraging local internet restrictions to capture overseas demand.
Source

Inside Amsterdam’s Welfare-AI Experiment

The Dutch capital’s attempt at bias-free social-benefits AI falls short, raising doubts about “fair” algorithmic policy.
Source

NVIDIA’s Industrial AI Cloud Goes Live
10 000 DGX B200 & RTX PRO GPUs in Germany kick off NVIDIA’s $50 T “physical AI” vision for manufacturing and robotics.
Source

AMD Announces New GPUs, Development Platform, and Rack Scale Architecture
At the Advancing AI 2025 event, AMD unveiled new GPUs, the ROCm 7.0 software stack, and the Helios rackscale system, aiming to challenge NVIDIA’s dominance in the AI and GPU markets.

Source

DeepL Deploys NVIDIA DGX SuperPOD with Grace Blackwell 200 Systems
DeepL announced the deployment of an NVIDIA DGX SuperPOD featuring DGX Grace Blackwell 200 systems, enabling the company to translate the entire internet in just over 18 days, a significant leap in AI hardware capability.

Source

Fujitsu and Japan Announce Plans for Zetta-Class Supercomputer

Fujitsu, in collaboration with Japan’s RIKEN research institute, has unveiled plans to design and build the world’s first “zetta-class” supercomputer. This next-generation system will be capable of performing one sextillion (10²¹) calculations per second—over 10 times faster than today’s most powerful exascale supercomputers.

Source

Intel Discontinues ARC A750 and A770 Alchemist GPUs
Intel officially discontinued its ARC A750 and A770 Alchemist GPUs, signaling a shift in its GPU strategy and possibly paving the way for new product lines.

Source

UK’s £2bn AI Infrastructure Program
The UK government’s £2 billion investment is aimed at expanding national AI infrastructure and boosting sovereign data center capabilities, positioning the UK as a global AI leader.
Source

Google Expands Sovereign Cloud for EU Compliance
Google Cloud has introduced new enhancements to its sovereign cloud portfolio, including expanded data boundary controls, security validation tools, and local partnerships in Europe. These updates address rising EU requirements for data residency and regulatory compliance, with new offerings such as User Data Shield and Google Cloud Dedicated in France and Germany.

Source

AWS Prepares Launch of European Sovereign Cloud
AWS is set to launch its European Sovereign Cloud, with a major showcase at the upcoming AWS Summit Hamburg 2025. The new cloud offering will provide advanced sovereignty controls and compliance features, enabling European organizations to retain full control over their data and meet stringent regulatory requirements.

Source

European Business Leaders Call for Sovereign AI Infrastructure
New research published in June 2025 indicates that nearly half of UK, German, and French business leaders now consider sovereign AI infrastructure essential for national competitiveness and data security, highlighting a strong industry shift toward localized and controlled data center solutions.
Source

Nearly Half of European Business Leaders See Sovereign AI Infrastructure as a Key Competitive Advantage

New research from Nscale reveals that almost 50% of senior decision-makers in the UK, France, and Norway view sovereign AI infrastructure as a major competitive differentiator by 2027. The study finds that European cloud providers are trusted more than their US counterparts, driven by concerns over compliance with EU data protection laws.
Source

Multiverse Computing Raises $215M for LLM Compression
Multiverse Computing announced a $215 million funding round to advance CompactifAI, a technology that compresses large language models by up to 95% while maintaining performance.

Source

Global AI Market Valued at Nearly $400 Billion, Set to Grow 5x by 2030
The global AI market is currently worth about $391–$400 billion and is projected to grow nearly fivefold to $1.81 trillion by 2030, reflecting a surge in both public and private investment worldwide.

Source

1997 – IBM’s Deep Blue Defeats Garri Kasparov (Rematch)
One year after their historic match, on June 18, 1997, IBM’s Deep Blue supercomputer convincingly beat Garri Kasparov in a rapid–play exhibition rematch, underscoring the rapid advance of chess engines.

2015 – DARPA Robotics Challenge Finals Kick Off
On June 18, 2015, the finals of the DARPA Robotics Challenge began at Fairplex in Pomona, California, where cutting-edge humanoid robots from around the world tackled disaster-response tasks under challenging real-world conditions.

2017 – Carnegie Mellon’s Libratus Tops Poker Pros
After its January victory, on June 18, 2017, CMU’s Libratus AI system completed a public “heads-up” no-limit Texas Hold’em challenge, cementing its status as the first AI to conclusively beat top human professionals in a game of imperfect information.

2020 – GPT-3 Beta Expanded to Academic Researchers
On June 18, 2020, OpenAI broadened access to its GPT-3 private beta, inviting select university labs and nonprofit research groups to begin exploring its unprecedented 175 billion-parameter model.

2022 – Tesla Introduces “Full Self-Driving” Beta to Early Access Users
Tesla rolled out its first widely-publicized “Full Self-Driving” beta on June 18, 2022, allowing hundreds of owners in North America to test advanced driver-assist features on public roads under guidance from Tesla’s neural-network stack.

Google Gemini 2.5 Flash-Lite
Generally available: fastest, most cost-efficient of the 2.5 family, supporting multimodal input, 1 M-token contexts, tool integration (Search, code).

Adobe Firefly Goes Mobile
iOS & Android release brings Generative Fill, Expand, video tools, and Imagen/Sora models to phones with Creative Cloud sync.

DeepSeek R1-0528 & FLUX.1 Kontext on Together AI
Now available via Serverless Inference API: DeepSeek adds reasoning & function-calling; FLUX.1 offers natural-language image editing.

MiniMax M1 Open Reasoner
Released June 17: 1 M-token context window, 80 k “thinking budget,” matched leading open models, trained in 3 weeks for $535 k using new CISPO RL algorithm.

NAD Flags Microsoft 365 Copilot Claims
June 16: Advertising watchdog urges clearer disclosures on Business Chat’s limitations after finding unsupported productivity percentages.

OpenAI–Microsoft Tensions Boil Over
June 16: OpenAI weighs antitrust action against Microsoft over the $3 B Windsurf deal and compute-access disputes amid restructuring.

Character AI’s Writing Evaluation Framework
New scoring system blends creative writing heuristics with objective metrics for conversational storytelling.

Gallup: Workplace AI Use Doubles
June 17 survey: 40 % of U.S. employees now use AI at work, led by tech roles (50 %), with frontline at 9 %.