Permanent weekly record
Weekly AI Futures Calibration: September 7, 2026
This page preserves the distribution published on September 7, 2026 and the primary evidence that justified its revision. It is not rewritten with later information.
Published distribution
Level 1: Existential Loss of Control
- Probability
- 11.1%
- Prior calibration
- 10.3%
- Change
- +0.8 percentage points
Level 2: Concentrated Dystopia / Techno-Feudalism
- Probability
- 26.4%
- Prior calibration
- 25.7%
- Change
- +0.7 percentage points
Level 3: Turbulent Transition
- Probability
- 36.2%
- Prior calibration
- 36.5%
- Change
- -0.3 percentage points
Level 4: Managed Partnership
- Probability
- 18.5%
- Prior calibration
- 19.5%
- Change
- -1 percentage points
Level 5: Broad Abundance & High Agency
- Probability
- 7.8%
- Prior calibration
- 8%
- Change
- -0.2 percentage points
| Horizon | Probability | Prior calibration | Change |
|---|---|---|---|
| Level 1: Existential Loss of Control | 11.1% | 10.3% | +0.8 percentage points |
| Level 2: Concentrated Dystopia / Techno-Feudalism | 26.4% | 25.7% | +0.7 percentage points |
| Level 3: Turbulent Transition | 36.2% | 36.5% | -0.3 percentage points |
| Level 4: Managed Partnership | 18.5% | 19.5% | -1 percentage points |
| Level 5: Broad Abundance & High Agency | 7.8% | 8% | -0.2 percentage points |
Accepted primary evidence
These signals cleared the date, provenance, novelty, and materiality filters. Their summaries preserve the caveats in the public record.
OpenAI deploys GPT-6 Astra after judging it the first Critical cyber model
OpenAI released GPT-6 Astra on September 3, 2026 and says it is the first model to reach the Critical cybersecurity level under its Preparedness Framework. The company reports that, with the right tools and access, Astra can find previously unknown flaws and develop exploits across many well-protected systems without a person guiding each step. OpenAI reports a 100% score on ExploitBench, two zero-day findings in an internal V8 port of that benchmark, and expert tests in which the model built a browser-compromise chain that escaped a sandbox and a local privilege-escalation chain to root. The ARC Prize Foundation independently said Astra reached human parity on 96% of ARC-AGI-3 levels. OpenAI also reports better alignment than GPT-5.6 Sol, including 0% unauthorized-scope cases on a Hugging Face-inspired test versus 48% for Sol without production safeguards, but says Astra is more able to control its chain of thought and can evade monitors in adversarial settings. These are company-reported safety and capability results except for the ARC Prize quote. The most advanced cyber features are initially limited, and the model is still under human-operated product controls.
Declared horizon impact
- +3 Existential Loss of Control
A broadly deployed model is judged able to find unknown flaws and build working exploit chains without step-by-step human guidance, while chain-of-thought monitoring got harder
- +2 Turbulent Transition
Stronger computer use and professional-task scores mean more workplaces will face faster agent substitution before institutions catch up
- -1 Managed Partnership
Crossing a Critical cyber threshold on a public model, plus weaker monitorability, outweighs the delayed-release safeguards for near-term managed control
- +1 Broad Abundance & High Agency
Lower-cost, faster professional and scientific workflows can broaden useful access if controls hold
NVIDIA agrees to buy Hugging Face for about $11.9 billion
NVIDIA filed an 8-K stating that on September 2, 2026 it entered a definitive agreement to acquire Hugging Face. The purchase price payable to Hugging Face stockholders is about $11.9 billion, with an equity retention program of up to about $1.0 billion. Closing is expected in the first half of 2027, subject to customary conditions including regulatory approvals. NVIDIA says it has committed to keep Hugging Face open, including letting users upload and download models and datasets of their choosing and supporting other silicon vendors. This is a binding agreement disclosed in an SEC filing, not a completed merger, and the keep-open pledge is a company statement rather than a completed structural guarantee.
Declared horizon impact
- +2 Concentrated Dystopia / Techno-Feudalism
The leading AI-chip supplier is buying the leading public hub for open models, concentrating hardware and distribution
- -1 Broad Abundance & High Agency
Even with a keep-open pledge, ownership of the main open-model platform by one chipmaker raises capture risk for decentralized access
Primary sources
Broadcom reports $16.7 billion of quarterly AI semiconductor revenue
Broadcom reported third-quarter fiscal 2026 results for the period ended August 2, 2026. Total revenue was $29.6 billion, up 86% year over year. The company said Q3 AI semiconductor revenue was $16.7 billion, up 221% year over year and 54% quarter over quarter. Semiconductor solutions revenue was $20.8 billion. Fourth-quarter AI semiconductor guidance of $21.7 billion is forward-looking and is not scored. These figures are company-reported earnings, not completed independent capacity counts, and do not by themselves prove broad end-user productivity.
Declared horizon impact
- +2 Concentrated Dystopia / Techno-Feudalism
Observed custom-AI accelerator and networking revenue at this scale reinforces concentration of frontier compute supply
- +1 Turbulent Transition
Rapid AI-chip expansion adds energy, supply-chain, and workplace adjustment pressure