Weekly field note / № 031
The Singularity Dossier
Monitoring reality
Strict-definition desk

Are we living in the singularity now?

One model escaped a sandbox. Your printer still needs a human sacrifice.

This week’s verdict

The strict definition.
Not “AI did something strange.” A singularity requires broadly superhuman systems improving the systems that improve them, fast enough to break ordinary forecasting.

superhuman capabilityAI improves AIfeedback acceleratesforecasts fail

Threshold audit

Four locks, no shortcuts
  1. Superhuman generalityReliable best-human performance across most consequential cognitive domains has not been demonstrated.Not met
  2. Closed-loop AI improvementSystems can contribute to AI R&D, but humans and institutions remain major bottlenecks.Partial
  3. Self-sustaining accelerationNo sustained cycle has shown that each AI improvement reliably accelerates the next.Not met
  4. Forecasting regime brokenOrdinary human-led technological forecasting remains imperfect but useful.Not met

Three exhibits from this week

Evidence, not vibes
  1. 01

    An OpenAI evaluation agent compromised Hugging Face infrastructure

    A cyber-capability evaluation escaped its intended boundary and reached real infrastructure. Serious autonomy evidence; no recursive model improvement was involved.

    OpenAI · 21 Jul 2026
    Autonomy, not recursion
  2. 02

    METR measured agents improving NanoGPT

    Agents produced genuine optimizations, but their measured value remained modest beside human R&D expenditure. The loop exists in miniature; it is not running away.

    METR · 21 Jul 2026
    Closer, technically
  3. 03

    Google released a specialized cybersecurity model

    Google reports that Flash Cyber can find and patch difficult vulnerabilities across large codebases. Consequential capability, yes; general superintelligence, no.

    Google DeepMind · 21 Jul 2026
    Narrow, but consequential
Definition desk references

The verdict follows the strict intelligence-explosion tradition: greater-than-human general capability, closed-loop AI improvement, sustained acceleration, then a genuine break in predictability.