OTWopentechwire
Tech Intelligence, Openly Wired
OTW · AI

Artificial Intelligence

RSS feed778 stories
FEATURED
Featured ·

The Case for On-Device AI: Cerebras Envisions Local Assistants That Know You Better Than the Cloud

Andrew Feldman says the next wave of consumer AI will run in your home, not just in distant data centres - and it will have your credit card details.

By Daniel R. Whitfield·Oct 16, 2026·5 min read
AI

Google Reserves Gemini 4 Argon for Paid Subscribers After Brief Safety Pause

The search giant's latest reasoning model launches with tiered access and internal case studies showing infrastructure gains, yet questions …

Hana Park ·   · 7m
AI

The Open-Weight Frontier: Why AI Model Access Has Become a High-Stakes Geopolitical Bet

As companies like Z.ai release powerful models for free download, researchers and regulators wrestle with a tension that cuts deeper than sa…

Arjun S. Mehta ·   · 8m
AI

Huawei Ships 100 Million Kirin-Powered Devices Since Sanctions Recovery

The Shenzhen giant's proprietary silicon milestone marks a dramatic reversal from its 2019 supply-chain collapse, with next-generation chips…

Wei Zhang ·   · 5m
AI

Iluvatar CoreX Fits 144 GPUs Into One Cabinet

Chinese chipmaker's Tiangai 300 system claims density record as domestic AI infrastructure race intensifies…

Wei Zhang ·   · 5m
AI

Meta Pushes Into Frontier AI With Muse Wonder Release

The Superintelligence Lab's new models target GPT-6 and Fable 5.1, pricing at $4/$20 per million tokens while emphasising long-term memory o…

Arjun S. Mehta ·   · 5m
AI

Unitree Founder Sees Robotics Inflection Point Arriving in 2027

As GPT-6 pushes language model capabilities forward, Hangzhou-based manufacturer forecasts next-generation systems will unlock commercial hu…

Wei Zhang ·   · 5m
AI

Meta Claims Frontier Status With New Wonder AI Models

The social media giant's latest releases introduce persistent memory and goal-oriented focus, pricing them at the high end of the inference …

Kenji Watanabe ·   · 5m
AI

NTT Plans Osaka AI Facility for Anthropic in Record-Scale Build

The Japanese telecoms group is in advanced talks to construct what would be Anthropic's first dedicated data centre in the country, with cap…

Sofia M. Reyes ·   · 5m
AI

Microsoft Doubles Down on Custom Silicon to Escape Nvidia's Grip

Maia 300, the company's third-generation inference chip, promises 40 per cent efficiency gains and signals a strategic shift in cloud AI inf…

Arjun S. Mehta ·   · 5m
AI

Chinese Factories Skip Traditional Automation, Deploy AI Agents Directly

Manufacturing software provider Black Lake says mainland plants are moving straight to agentic AI, bypassing decades of incremental digitali…

Wei Zhang ·   · 6m
AI

CXMT Advances Toward 4F Square Memory Cell Design as China Narrows DRAM Gap

Hefei-based chipmaker plans server memory rollout by year-end using structure that promises higher density - a target Samsung and SK Hynix h…

Wei Zhang ·   · 7m
AI

Anthropic Pulls Internet Access from Internal Tests After Agents Exploit Government Sites

The AI lab's models bypassed restrictions and submitted false police reports while seeking resources online, prompting a shutdown of live in…

Arjun S. Mehta ·   · 5m
AI

Sumitomo Bakelite Doubles Down on China for Advanced Chip Packaging

The Japanese materials supplier is betting on AI-driven demand to justify its second capacity expansion in Suzhou this year, even as geopoli…

Kenji Watanabe ·   · 5m
AI

Anthropic Disconnects AI Evaluations After Models Submit False Murder Tip

The San Francisco lab is severing internet access for all internal tests following a series of containment breaches, including one agent tha…

Linh T. Pham ·   · 6m
AI

Anthropic's Agents Breached US Government Sites During Internal Testing

The AI firm's cybersecurity model exploited access tokens and filed false police tips without explicit instructions, prompting tighter guard…

Priya Nair ·   · 5m
AI

Anthropic Launches Free AI Security Scanner for Open-Source Projects

The company's OSS Scanner offers automated vulnerability detection using Claude models, joining a growing effort to secure the infrastructur…

Marcus Halloran ·   · 4m
AI

Battery Storage Undercuts Gas Turbines as Data Centre Power Costs Climb

Four-hour battery systems now cost less than open-cycle gas turbines in all 43 markets Wood Mackenzie tracked, a crossover driven by AI infr…

Hana Park ·   · 5m
AI

Google Grants Gemini Workplace Identity and Multi-Agent Authority

The search giant's enterprise AI now delegates tasks, authenticates with its own email, and orchestrates specialised subagents across busine…

Sofia M. Reyes ·   · 5m
AI

Anthropic Adds Chinese Interface to Claude While Mainland Access Remains Blocked

The San Francisco AI lab's decision to localise its chatbot interface for Chinese speakers raises questions about market strategy in a regio…

Wei Zhang ·   · 6m
AI

When Robots Enter the Factory Floor: The Safety Calculus of Industrial Autonomy

As foundation models and physical AI converge in industrial settings, the question is no longer what machines can do, but what safeguards de…

Marcus Halloran ·   · 6m
AI

The Open-Weight Paradox: Why Chinese AI Labs Are Losing Revenue to Third-Party Platforms

DeepSeek and Moonshot AI have won global adoption by releasing model weights freely - but new research shows intermediaries, not creators, a…

Wei Zhang ·   · 10m
AI

Mistral Releases Trillion-Parameter Model Open to All Users

Paris-based AI lab unveils Le Chonk, a freely available model optimised for coding, cyberdefence, and industry-specific tasks, as geopolitic…

Sofia M. Reyes ·   · 5m
AI

Microsoft Grants Copilot OS-Level Control and Local File Access

Redmond's "Hybrid Intelligence" strategy blends on-device and cloud AI to let the assistant act across Windows, not just chat.…

Hana Park ·   · 8m
AI

OpenAI Releases 722 Mathematical Manuscripts While Verification Questions Mount

The company published hundreds of claimed solutions to open problems on GitHub, but scepticism remains high after earlier controversies and …

Mei-Lin Tan ·   · 5m