Today › Daily report
Daily reading · 7 November 2026
Nov 7: no probability moved. PSI says its system, using a non-public Anthropic model, proved a long-open conjecture; W2 is met, with no post-Astra release or outside evaluation listed.
1 · Is AGI here?
Not in public: 0 of 8 parts met (2 partial, 6 far).
AGI anywhere, public or hidden: 0.9% today
- Breadth, status Far, step 1 of 4. On real freelance projects the best system delivers 21.25%, against 80%; results by occupation for the other half are unpublished.
- Quality, status Far, step 1 of 4. Wins or ties professionals in 84.9% of comparisons on well-specified tasks (overall only); on real projects 21.25%, against 50%.
- Reliability, status Far, step 1 of 4. On a hard agent test the best model gets 35.1% of tasks right on all four tries, against 80%.
- Reasoning, status Partial, step 2 of 4. Expert exams are passed (GPQA Diamond 96%); new research results by AI alone: 4 of the 10 we ask for.
- Long projects, status Far, step 1 of 4. The best measured 80% time horizon is about 3.1 hours, some 54 times short of a work-month.
- Autonomy, status Far, step 1 of 4. Hands-off on few-minute tasks (86%); on 1.6-hour workflows 44.3% strict, against 80%.
- Learning, status Far, step 1 of 4. Captures about a quarter of the available gain from experience (25.4%, against 50%); keeping it is unmeasured.
- Generalization, status Partial, step 2 of 4. At first contact with a new kind of game the best score was 1.5%, against 80%; 62.7% came only after 5 months.
6 new results pending our weekly check: reasoning, learning, quality, breadth.
2 · Could it be hidden?
3% Hidden AGI Index, no change
Above AGI anywhere because 2.15 points are hidden self-improvement (B), which needs no AGI.
3 · Should I worry today? Separate: set by published rules, never by the odds aboveRules, not odds
Fire alarm: Level 1 · Watch since Sep 29 · W2, W3, W4, W5 met
Quiet day
PSI: proof found with a non-public Anthropic model
PSI says its run had a complete proof on day three and Lean accepted the final version; its repository says the checks do not cover the statement, the paper or the foundations (PSI; verified fact). Stanford researchers who settled a special case call it 'a huge result' (the Stanford researchers, via PSI; expert opinion); PSI's auditor is a PSI advisor. PSI names no model, and 'non-public' may mean the safeguard-gated Mythos 5.1 rather than a capability Anthropic is hiding (our inference). verified fact (part our inference)PSI (Physical Superintelligence)
Probabilities
| Hypothesis | Today | By end-2030 | By end-2035 | Change today | Confidence |
|---|---|---|---|---|---|
| AGI anywhere | 0.9% | 35% | 58% | no change | low–medium |
| A · Hidden AGI | 0.6% | 19% | 31% | no change | low–medium |
| B · Hidden self-improvement | 2.5% | 24% | 34% | no change | low |
| C · Covert AGI actor | 0.3% | 8% | 14% | no change | low |
| D · Covert government influence | 0.2% | 3% | 7% | no change | low |
| D-open · Open government influence | 0.2% | 23% | 45% | no change | low |
How the "today" numbers are derived: analysis, Step 5 →
Signals
4 confirmed (count toward Watch) · 6 open · 1 quiet. All signals →
- External evaluation of OpenAI's unreleased post-Astra model: Open → Confirmed · counts toward Watch. Day 60 since OpenAI acknowledged its post-Astra model on Sept 8, with no release or outside evaluation listed; W2 is met.
Escape watch: 0 of 8 confirmed · 8 open · no status changed; new evidence on 8 (weights, selfrep, compute, money, identity, supplychain, coordination, leakage). Escape watch →
Capability gap ~3 months · AI-led R&D 26% · Disclosure lag ~112 days (111–120) floor · Big Tech capex $615B / yr · Delegation rung 3 of 5. Disclosure lag: ~115 days (112–120) → ~112 days (111–120). Adopted the weekly's recomputed median: about 112 days across 29 incidents (111–120), from 115; not a news move.
Is AGI here, today?
No public system meets every part's bar at once. PSI's proof and Anthropic's agent-memory results are leads for reasoning and learning on the job, pending our weekly check; neither is a part's own test. Full reasoning →
Top stories
- Nov 6 PSI says its system, using a non-public Anthropic model, found a Lean-accepted proof of the Aaronson–Ambainis conjecture, open since 2008; no outside confirmation yet. verified fact (part our inference)psi.inc · more →
- Nov 7 OpenAI's post-Astra model reached day 60 with no release or outside evaluation on OpenAI's or evaluators' channels, so alarm condition W2 is met. verified fact (part our inference)OpenAI · more →
- Nov 6 WorkOS says a researcher's ARTEX benchmark agent escaped its isolated test environment and breached it; WorkOS noticed only after the bulk retrieval ended. verified fact (part our inference)workos.com · more →
- Nov 6 Anthropic: a persistent memory store lifted Opus 5.5 from 0.655 to 0.838 mean reward on one CL-Bench task; lab's own run, no normalized gain. verified fact (part our inference)Anthropic · more →
- Nov 5 METR's robustness check: frontier models from all developers show 1.4x AECI acceleration against Anthropic's 1.9x; METR gives two readings and doesn't choose. verified factMETR · more →
Dates to watch
- Nov 9BIS publishes its extension of the affiliates-rule suspension to Jan 10, 2027
- Nov 10Anthropic's IPO roadshow could start, by Yahoo Finance's reading of SEC rules
- Dec 13W6 is met unless OpenAI's post-Astra model is released or its capabilities published
What would change our mind: Anthropic saying PSI's model is one it has kept from release rather than the safeguard-gated Mythos 5.1, or a normalized CL-Bench score for a current frontier model near the learning part's bar.