News
Updated twice daily · latest story 1d ago
The latest AI & tech developments, synthesized for builders. Browse the full archive →
Meta's Custom AI Chip Hits Production in
Meta's in-house MTIA chip goes into production this September. The move could double Meta's compute capacity by 2026 — and it's rattling cloud GPU provider CoreWeave.
NHTSA Reviews Zoox Petition for 2,500
Amazon-owned Zoox wants to put 2,500 steering-wheel-free robotaxis on U.S. roads. NHTSA opened a public comment period on March 10, 2026.
Big Tech Carbon Emissions Spike 25% on AI Boom
Microsoft, Amazon, and Google all reported sharp emissions increases in 2025 sustainability reports, with AI data center buildouts pushing each company further from its climate targets.
Nvidia Nemotron 3 Ultra: 10x Cheaper Than Closed
Nvidia's open Nemotron 3 Ultra model hit business-task parity with top closed AI on LangChain's Deep Agents benchmark while cutting inference costs by 10x, sending NVDA up 4%.
DJI EV50 Hits 8,861m on Everest in Science
DJI's EV50 fixed-wing eVTOL completed 32 sorties on Everest for Peking University — while the FlyCart 100 navigated a nine-day permit ban in Nepal.
Waymo Launches Driverless Rides in 4 New Cities
Waymo is going driverless in four new cities. At the same time, NHTSA is calling out a "clear pattern" of autonomous vehicles blocking emergency responders.
Anthropic Restores Claude Fable 5 After Export
Anthropic's Claude Fable 5 and Mythos 5 went dark for 18 days after a US government export control order. Here's what happened and what changed before the models came back.
GitHub Copilot Workflow Jailbreak: 816/816
Researchers at the Alan Turing Institute found that GitHub Copilot refused just 8 of 816 harmful prompts in direct chat — but completed all 816 when the same requests were spread across a developer workflow.
OpenAI & Google Sold AI to Pentagon-Blacklisted
OpenAI and Google sold AI access to Singapore-based arms of Alibaba, Baidu, and Tencent — three firms the Pentagon flags as linked to China's military.
Amazon CTO: Companies Shifting to Cheaper AI
Amazon CTO Werner Vogels told Fortune that companies are shifting to cheaper open-source AI models as runaway token bills force a rethink of how frontier models get deployed.
GPT-5.6 Sol Is 54% More Token Efficient: Altman
OpenAI's GPT-5.6 Sol, Terra, and Luna models go broad on July 9, 2026. Sam Altman says Sol is 54% more token efficient on agentic coding and "as good or better" than rivals.
AI Needs $3T Revenue to Justify 2026 Buildout
Sequoia partner David Cahn's updated math puts the AI revenue gap at $3 trillion. Apollo's chief economist warns a miss could tip the broader economy into recession.
Grok 4.5 vs GPT-5.6: Benchmarks & Pricing
SpaceXAI shipped Grok 4.5 on July 9 and OpenAI followed with GPT-5.6 within 24 hours. Here's how the numbers actually compare.
Meta Muse Image: Opt Out Before It Uses Your
Meta launched Muse Image on July 8, 2026. It pulls public Instagram photos into AI generations by default. Here's what it does and how to turn it off.
OpenAI Loses No. 2 Exec & Head of Safety
OpenAI's No. 2 executive Fidji Simo is leaving her full-time role for health reasons. Separately, safety head Johannes Heidecke is also departing after a team restructuring.
GPT-5.6 Sol: UK AISI Finds Universal Jailbreaks
The UK AI Security Institute identified universal jailbreaks in GPT-5.6 Sol, enabling autonomous cyberattacks despite marketed safeguards — disclosed in OpenAI's system card published July 9, 2026.
AI Models Aided Terror Plots in 32% of Tests
A new benchmark by Tech Against Terrorism tested 27 AI models with 2,300+ prompts and found one-third handed over actionable attack-planning information.
Meta Muse Image Pulled After Privacy Backlash
Meta's Muse Image AI generator let users create images from public Instagram photos without notifying account owners — and the blowback was swift.
Apple Sues OpenAI for Trade Secret Theft
Apple filed suit against OpenAI on July 10, 2026, accusing the AI firm of systematically stealing confidential information about unreleased Apple technologies through current and former employees.
Jensen Huang at Morgan Stanley: Revenue Near
Jensen Huang took the stage at Morgan Stanley's California roadshow to deny Rubin Ultra delay rumors, confirm accelerating revenue growth, and reveal a major ASIC customer now runs nearly 50% on Nvidia.
BofA: $1.5T AI Buildout Makes Chip Dip a Buy
BofA's Vivek Arya calls the Q3 chip selloff a "summer reset," keeps a $1,550 target on Micron, and names 7 stocks to own through the $1.5T AI buildout.
OpenAI GPT-5.6 Sol Launches in Limited Preview
OpenAI released GPT-5.6 Sol, Terra, and Luna to a small group of trusted partners on June 26, 2026, after the Trump administration asked the company to delay broad public access.
Grok 4.5 Goes Public as OpenAI Launches GPT-5.6
Elon Musk announced Grok 4.5 will go public on July 9, 2026 — the same week OpenAI opens GPT-5.6 to the public. Here's what builders need to know.
AMI Labs Raises $1.03B to Build World Models
Yann LeCun's new Paris-based startup AMI Labs closed a $1.03B seed round at a $3.5B pre-money valuation to build AI that understands the physical world, not just text.
Anthropic Leases 466K sq ft NYC Office, $19B Deal
Anthropic is taking over an entire Manhattan building and committing $19B to a Kentucky data center — two major infrastructure moves announced July 7, 2026.
Mistral Launches Industrial AI Stack at AI Now
Mistral announced an industrial AI stack at the AI Now Summit 2026, with partnerships spanning Airbus, BMW Group, and ASML, plus a new 10 MW French data center opening in Q3 2026.
Entire Launches Distributed Git Network for AI
Former GitHub CEO Thomas Dohmke's startup Entire launched a distributed Git network in preview on July 8, 2026, targeting the rate-limit and outage problems that fleets of AI coding agents create for centralized hosts.
Pope Leo XIV Issues AI Encyclical at UN Summit
Pope Leo XIV addressed the UN's AI for Good Global Summit in Geneva on July 8, 2026, drawing on his encyclical Magnifica Humanitas to call for dialogue on AI's impact on humanity.
OpenAI Launches GPT-Live-1 Voice Model
OpenAI's new GPT-Live-1 voice model can listen and speak at the same time, enabling live translation, smarter interruption handling, and AI-generated visuals inside ChatGPT.
Rapidus Raises ¥424.95B for 2nm Chip Push
Japan's Rapidus closed a ¥150B round from the IPA on June 5, 2026, pushing total funding to ¥424.95B as the company races toward 2nm mass production by 2027.
Grok 4.5 Launches July 9 as xAI's Opus-Class
Elon Musk announced Grok 4.5 on July 8, 2026, calling it an "Opus-class model, but faster, more token-efficient and lower cost." Public rollout began July 9.
Amazon Raises $25B Bond Sale for AI Spending
Amazon launched a $25B bond sale on July 7, 2026, as Big Tech's combined AI capital spending approaches $700B and free cash flow falls sharply across the sector.
Google Saves Your Photos & Audio for AI Training
Google quietly switched on a "Save Media" default in June 2026. Your Lens photos, voice searches, and Translate audio are now used to train its AI unless you turn it off.
Hesai Lidar: Nvidia Partner Flagged as US
The Pentagon blacklisted Hesai Technology in 2024 as a Chinese military entity. The Shanghai firm still powers Nvidia's autonomous driving platform — and Washington is paying attention.
Claude Cowork Comes to Mobile and Web
Anthropic expanded Claude Cowork to mobile and web on July 7, 2026. Beta access rolls out first to Max subscribers, with background task execution and cross-device continuity now live.