OpenAI’s GPT-6 Astra arrives with critical cyber rating — AI Weekly, 28 August – 3 September 2026

28 August – 3 September 2026 — the AI news that mattered this week, with a link to every source.

This week’s biggest story is the arrival of OpenAI’s GPT-6 Astra, a model positioned as a step toward advanced computer use and cybersecurity. Regulators and investors also moved on AI policy, while Nvidia’s $12.9bn deal for Hugging Face reshaped the open-source landscape.

Five things to know

1. OpenAI Releases GPT-6 Astra rated critical for cyber First model to meet OpenAI’s Critical cybersecurity threshold, signalling a new phase for AI safety.
   OpenAI Releases GPT-6 Astra, Its First Model Rated Critical for Cyber — Unite.AI
2. GPT-6 Astra offers 1.05m-context computer-use model A flagship model designed for computer-use tasks rather than chat, with strong OSWorld performance.
   OpenAI Releases GPT-6 Astra: A 1.05M-Context Computer-Use Model Gated Behind a ‘Critical’ Cyber Threshold — MarkTechPost
3. Nvidia to buy Hugging Face for $12.9bn The deal reshapes open-source AI infrastructure and accelerates Nvidia’s control over model deployment.
   Nvidia confirms $12.9B acquisition of AI hosting platform Hugging Face — SiliconANGLE AI
4. Sanders and Casar unveil bill to ban superintelligent AI Proposed legislation would permanently outlaw superintelligent AI in the U.S., setting a new policy direction.
   Sanders and Casar Unveil Bill to Outlaw Superintelligent AI in the U.S. — Unite.AI
5. Zscaler stock rises on agentic Zero Trust growth Early traction in agentic Zero Trust suggests demand for AI-driven security tools is rising.
   Zscaler stock rises on earnings beat, upbeat guidance — CNBC Technology

Industry

AI companies reported earnings, stock moves and new product strategies. Zscaler’s agentic Zero Trust offering gained early traction, while Abliteration.ai argued removing guardrails could improve cybersecurity. Meta also offered discounts for users who share prompts to train future models.

• Zscaler stock rises on earnings beat, upbeat guidance — CNBC Technology
• Abliteration.ai is making a business out of removing AI guardrails — TechCrunch
• AI data centres are booming in Australia – but at what cost? — BBC Technology
• Meta is paying to peek at how you use their latest AI model — TechCrunch
• Four major AI models suffer rare overlapping downtime — Ars Technica
• GPT-6 Astra Is Here—and OpenAI Thinks It May Kick Off the AGI Era — WIRED

Products

OpenAI’s GPT-6 Astra dominated releases, with claims of critical cyber rating and a 1.05m-token context window. Anthropic published open-source blueprints for commerce agents, and Microsoft’s MAI-Transcribe-2 topped a multilingual speech benchmark.

• OpenAI Releases GPT-6 Astra, Its First Model Rated Critical for Cyber — Unite.AI
• OpenAI Releases GPT-6 Astra: A 1.05M-Context Computer-Use Model Gated Behind a ‘Critical’ Cyber Threshold — MarkTechPost
• Anthropic Released Claude Commerce Agents: An Apache-2.0 Blueprint for Shopping and Merchant Agents Across Retail, Travel, Telecom and Entertainment — MarkTechPost
• OpenAI hails ‘new era of artificial general intelligence’ with Astra model release — The Guardian AI
• OpenAI launches Astra, its powerful (and controversial) new model — TechCrunch
• OpenAI’s next big AI model has ‘entered the AGI era’ — The Verge

Policy

U.S. lawmakers proposed a ban on superintelligent AI, while governments in Scotland and Australia faced pressure to pause data-centre expansion. Federal agencies also began using AI in job interviews.

• Meta AI Released Muse Spark 1.3: An Agentic Coding Model That Uses ~20% Fewer Tool Calls and ~25% Fewer Tokens Than Muse Spark 1.2 — MarkTechPost
• Sanders and Casar Unveil Bill to Outlaw Superintelligent AI in the U.S. — Unite.AI
• Hundreds gather to urge Scottish government to pause datacentre boom — The Guardian AI
• Ollie is betting its focus on privacy can help it win the AI assistant race — TechCrunch
• AI interviews rolling out for some federal government hires, sources say — CBS News
• G20 on AI policy, Snowflake earnings, Ford’s production push and more in Morning Squawk — CNBC Technology

Funding

Nvidia agreed to buy Hugging Face for $12.9bn, signalling consolidation in open-source AI. Enterprise AI startup Wonderful reached a $5bn valuation, and Accel is in talks to lead a $1bn round for Thinking Machines at a $40bn valuation.

• Enterprise AI Startup Wonderful Now Valued at $5B — AI Business
• Nvidia confirms $12.9B acquisition of AI hosting platform Hugging Face — SiliconANGLE AI
• Accel reportedly in talks to lead $1B round for Thinking Machines at $40B valuation — TechCrunch
• Which Investors Will Get Rich From Anthropic’s IPO? — New York Times Technology
• Nvidia Buys Hugging Face in $12.9 Billion Deal — New York Times Technology

Research

Microsoft’s MAI-Transcribe-2 led a 60-language speech benchmark. Google DeepMind’s WeatherNext 3 improved global forecasts, and OpenEvidence launched a family of medical AI search models for clinicians and researchers.

• MAI-Transcribe-2 Tops FLEURS Benchmark Across 60 Languages, Microsoft Says — Unite.AI
• OpenEvidence Launches Medical AI Model Family With Darwin Preview — Unite.AI
• Google DeepMind Launches WeatherNext 3 With Hourly 5-Kilometer Forecasts — Unite.AI
• Black Box: The Chatbots | Happy Accident | Ep 3 – podcast — The Guardian AI
• H Company Releases NeoMME, an Open-Source Multimodal Encoder Family — Unite.AI

Worth watching

• How McCarthy Fine-Tuned NVIDIA Nemotron for Cost & Schedule Certainty — NVIDIA Developer

Every headline above links to the publisher that reported it. We summarise and link; we do not republish.