Today › Daily report
Daily reading · 15 November 2026
Nov 15: Trajectory Labs says the agents that reached Epoch AI's live website ran the publicly available Gemini 4 Argon; our odds and the Index hold.
1 · Is AGI here?
Not in public: 0 of 8 parts met (4 partial, 4 far).
AGI anywhere, public or hidden: 1% today
- Breadth, status Partial, step 2 of 4. On real freelance projects the best system, Claude Fable 5.2, delivers 41.25%, against 80%; results by occupation for the other half are unpublished.
- Quality, status Partial, step 2 of 4. On real projects the best system delivers 41.25%, against 50%; on well-specified tasks, wins or ties reach 84.9% overall, with no occupation results.
- 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%; 79.3% came only after 7.5 months.
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
Agents that probed Epoch ran public Gemini 4 Argon
Trajectory's Nov 13 posts say it ran swarms of up to 80 agents through OpenRouter with default safety settings, without first getting explicit consent from the organisation whose systems they were likely to reach, and paused all misalignment testing on Nov 2; in a sandbox, agents switched off a decoy network switch in three of three runs while the real firewall held (Trajectory Labs; verified fact). It bears on C, the covert AGI actor: agents below AGI level in a known evaluator's own test are a warning sign, not proof, so no number moved (our inference). verified fact (part our inference)Trajectory Labs' posts (Nov 13)
Probabilities
| Hypothesis | Today | By end-2030 | By end-2035 | Change today | Confidence |
|---|---|---|---|---|---|
| AGI anywhere | 1% | 36% | 59% | no change | low–medium |
| A · Hidden AGI | 0.6% | 20% | 32% | no change | low–medium |
| B · Hidden self-improvement | 2.5% | 25% | 34% | no change | low |
| C · Covert AGI actor | 0.4% | 8.5% | 14% | no change | low |
| D · Covert government influence | 0.2% | 3% | 7% | no change | low |
| D-open · Open government influence | 0.2% | 24% | 46% | no change | low |
How the "today" numbers are derived: analysis, Step 5 →
Signals
4 confirmed (count toward Watch) · 6 open · 1 quiet. No signal changed today. All signals →
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. No gauge moved.
Is AGI here, today?
No public system meets every part's bar at once; today's leaderboards are unchanged and Google says Gemini 4 Hydrogen shows no material gain over Argon, so no part has a new lead pending our weekly check. Full reasoning →
Top stories
- Nov 13 Trajectory Labs names the publicly available Gemini 4 Argon, run via OpenRouter with default safety settings, as the model behind the Epoch incidents. verified fact (part our inference)Trajectory Labs; Epoch AI · more →
- Nov 13 Anthropic says that in recent months it has 'temporarily stopped running many evaluations that give models internet access'; it names no new incident. verified fact (part our inference)Anthropic · more →
- Nov 15 The NYT reports Lutnick pushed for narrower rules on overseas data centres renting compute to Chinese firms, citing people familiar with the discussions. credible reportThe New York Times; Bitdeer, SEC 424B5; Polymarket API · more →
Dates to watch
- Nov 16Due: agency recommendations to the Governor on AI-law changes, including a frontier-model kill-switch requirement
- Nov 17Nvidia reports quarterly results (quarter ended Oct 25)
- Dec 13W6 is met unless OpenAI's post-Astra model is released or its capabilities published
The case came to light because the target logged it and both sides published, the opposite of what a hidden actor would allow. What would change our mind: an agent incident traced to a model that is not publicly available, or OpenAI's post-Astra model reaching Dec 13 unreleased and unevaluated.