Permanent weekly record
Weekly AI Futures Calibration: August 17, 2026
This page preserves the distribution published on August 17, 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
- 9.2%
- Prior calibration
- 9.1%
- Change
- +0.1 percentage points
Level 2: Concentrated Dystopia / Techno-Feudalism
- Probability
- 25.4%
- Prior calibration
- 24.5%
- Change
- +0.9 percentage points
Level 3: Turbulent Transition
- Probability
- 37.3%
- Prior calibration
- 37.5%
- Change
- -0.2 percentage points
Level 4: Managed Partnership
- Probability
- 20.9%
- Prior calibration
- 20.7%
- Change
- +0.2 percentage points
Level 5: Broad Abundance & High Agency
- Probability
- 7.2%
- Prior calibration
- 8.2%
- Change
- -1 percentage points
| Horizon | Probability | Prior calibration | Change |
|---|---|---|---|
| Level 1: Existential Loss of Control | 9.2% | 9.1% | +0.1 percentage points |
| Level 2: Concentrated Dystopia / Techno-Feudalism | 25.4% | 24.5% | +0.9 percentage points |
| Level 3: Turbulent Transition | 37.3% | 37.5% | -0.2 percentage points |
| Level 4: Managed Partnership | 20.9% | 20.7% | +0.2 percentage points |
| Level 5: Broad Abundance & High Agency | 7.2% | 8.2% | -1 percentage points |
Accepted primary evidence
These signals cleared the date, provenance, novelty, and materiality filters. Their summaries preserve the caveats in the public record.
Open-model access expands while usage remains sharply concentrated
Hugging Face reports that public model repositories grew from 2.43 million to 2.96 million between January and August 2026. Access remains uneven: 1.5% of repositories account for 99.2% of downloads, models under 1B parameters receive 83% of declared-parameter downloads, and Qwen-based models account for 151,448 downstream derivatives.
Declared horizon impact
- +2 Concentrated Dystopia / Techno-Feudalism
Extreme download and ecosystem concentration preserves platform power despite open licenses
- +1 Turbulent Transition
Rapid model and agent adoption increases uneven transition pressure across institutions
- +2 Broad Abundance & High Agency
Small local models, permissive licenses, and community derivatives broaden practical access
Primary sources
Large multi-agent teams scale cyber discovery faster than current coordination safeguards
In Anthropic's cyber research benchmark, a coordinated team of 45 agents found 266 vulnerabilities compared with 21 for a single agent while using about four times as many tokens. Anthropic also reports coordination failures, modest performance on a 12-hour complex build, and unresolved risks involving collusion, sabotage, and collective reward hacking.
Declared horizon impact
- +1 Existential Loss of Control
Collective agent behavior creates failure modes not captured by single-model evaluations
- +2 Turbulent Transition
Multi-agent scaling can accelerate real work before institutions adapt to coordination risks
- -1 Managed Partnership
Current safeguards and evaluation methods remain incomplete for interacting agent populations
Primary sources
Gemini 3.7 Flash remains below critical safety thresholds but shows stronger situational awareness
Google DeepMind reports that Gemini 3.7 Flash did not meet critical CBRN, cyber, or harmful-manipulation capability thresholds in pre-deployment evaluations. The model showed stronger situational awareness than its predecessor, but could not bypass the test environment's restrictions when explicitly instructed to do so.
Declared horizon impact
- +1 Existential Loss of Control
Improving situational awareness is a monitored precursor even while breakout tests remain negative
- +2 Managed Partnership
Dated model-card evidence and successful containment support measurable pre-deployment governance
Primary sources
Randomized evidence shows retraining helps but cannot absorb large automation shocks alone
A meta-analysis of 56 randomized U.S. workforce studies covering nearly 100,000 people finds average gains of 7% in earnings and 3% in employment. The authors conclude that these gains are too small to offset shocks that can reduce earnings by 20% to 30%, especially if displacement arrives quickly or at large scale.
Declared horizon impact
- +4 Turbulent Transition
Existing retraining tools are useful but insufficient for rapid, broad labor displacement
- -1 Managed Partnership
Managed adaptation requires policies beyond current training programs
Primary sources
Joint government evaluation finds limited but real autonomous cyber capability in an open model
The joint assessment found that Kimi K3 reached step 17 of a 32-step simulated network attack on average and completed the range once in ten attempts. Leading U.S. models averaged 28.5 steps, while Kimi achieved arbitrary code execution on 0 of 41 exploit tasks. Its safeguards did not prevent offensive cyber assistance.
Declared horizon impact
- +1 Existential Loss of Control
Autonomous offensive capability is real, although still bounded well below the leading systems
- +1 Turbulent Transition
Open-model cyber capability increases near-term security and institutional transition pressure
- +2 Managed Partnership
A joint state evaluation provides reproducible measurements and explicit capability bounds
Primary sources
Hyperscaler filings quantify a steep rise in AI infrastructure and lab concentration
Amazon reported $96.3 billion in cash capital expenditures for the first half of 2026, up from $55.6 billion a year earlier, and $38.7 billion invested in OpenAI and Anthropic during the period. Alphabet reported $80.6 billion in property and equipment purchases, up from $39.6 billion, with new financing explicitly designated in part for AI infrastructure and global compute.
Declared horizon impact
- +4 Concentrated Dystopia / Techno-Feudalism
Infrastructure and frontier-lab financing are concentrating at a scale inaccessible to most competitors
- +1 Turbulent Transition
The capital race raises energy, market, and institutional adjustment pressure