🤖 AI News Top 10
September 8, 2026
infrastructuur
🔥 HOOG
Anthropic reportedly signs $517 billion in compute deals
Anthropic has reportedly signed $517 billion in compute contracts, shortly after CEO Dario Amodei warned rivals about reckless risk. It is one of the largest AI infrastructure investments ever, underlining the extreme capital intensity of frontier AI and Anthropic’s aggressive expansion.
agents
🔥 HOOG
GPT-6 Astra beat Portal start to finish without human help
GPT-6 Astra completed the puzzle game Portal entirely on its own in under 24 hours, with no human assistance. It is a striking demonstration of autonomous agent capabilities in complex 3D environments — a concrete milestone for long-horizon, self-directed AI tasks.
onderzoek
🔥 HOOG
AI-designed drug appears to turn back the body’s biological clock
An AI-designed drug appears to reverse the body’s biological clock in an early clinical trial. It is a promising breakthrough at the intersection of AI-driven drug discovery and longevity research, with potentially far-reaching implications for pharma.
enterprise
📌 MEDIUM
NavigateAI targets construction’s labor crunch with AI copilots
Eric Wu (founder of Opendoor) is building NavigateAI, which gives construction workers real-time, hands-free guidance via smartphones and Meta’s AI glasses, backed by $25 million from Elad Gil and Khosla Ventures. It shows AI penetrating traditional sectors and the rise of AI wearables on the job site.
cybersecurity
🔥 HOOG
PEEP turns Chrome and Edge into post-compromise backdoors
Researchers have revealed PEEP, an advanced Chromium-based post-exploitation toolkit that masquerades as a bookmarks extension and turns Chrome/Edge into backdoors for host command execution. It is a relevant AI/cybersecurity threat touching browser security and post-compromise detection.
governance
📌 MEDIUM
Matt Clifford to leave ARIA before Anthropic role becomes a ‘distraction’
Matt Clifford is leaving the UK’s ARIA (Advanced Research and Invention Agency) before his role at Anthropic becomes a conflict of interest. MPs welcomed the move but demanded clarity on how the conflict could arise — illustrating the growing governance tension between public AI research roles and commercial AI firms.
cybersecurity
📌 MEDIUM
Fake IT calls target executives in Microsoft 365 data theft
Threat hunters have revealed a widespread data-theft and extortion cluster targeting Microsoft 365 and other SaaS services via IT help-desk vishing, adversary-in-the-middle token theft, and residential-proxy logins. It is a growing AI-assisted social-engineering threat against executives and SaaS environments.
hardware
📌 MEDIUM
2027 Android phones could boost gaming graphics and AI with Arm’s new chip tech
Arm’s new chip technology promises AI-based upscaling and frame-rate improvement for mobile gaming, with better performance and energy efficiency for AI tasks on Android phones from 2027. AI acceleration is shifting to edge/mobile hardware, relevant for on-device AI.
hardware
📌 MEDIUM
First Xiaomi, then the world: Arm’s neural rendering boost for phone gaming
The Xiaomi 18 Fold launches in China with an Arm Mali G2-Ultra NX GPU in the custom Xring O3 chip, after five years of Arm’s neural rendering development. Neural rendering (AI graphics) is reaching consumer hardware — a step toward AI-native GPUs.
hardware
📌 MEDIUM
DRAM contract prices to grow only 13-18% in Q3 as PC buyers resist funding AI
DRAM contract prices are forecast to grow only 13-18% in Q3, as PC buyers are reluctant to keep funding the AI revolution; only essential refreshes are still happening. It is a signal of AI infrastructure saturation and price pressure in the memory market.
📈 Key Trends
- • Frontier AI is becoming extraordinarily capital-intensive — Anthropic’s $517B compute deals dwarf previous infrastructure bets.
- • Autonomous agents are crossing real-world thresholds, from beating Portal unaided to hands-free construction copilots.
- • AI is moving into drug discovery and longevity, with early clinical signals of biological-clock reversal.
- • AI-assisted social engineering and browser-based post-compromise toolkits are a growing security concern.
- • AI acceleration is shifting to the edge — neural rendering and on-device AI are reaching consumer hardware.
- • Governance tension between public AI research roles and commercial firms is intensifying.
Wat vond je van dit bericht?
