Today › Daily report
Daily reading · 24 October 2026
Oct 24: OpenAI says frontier training resumed Oct 1 and its red-team agents chained sandbox escapes in sanctioned tests; A and D rose for time passing, not this.
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 On real freelance projects the best system delivers 21.25%, against 80%; results by occupation for the other half are unpublished.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.
2 new results pending our weekly check: quality, reliability.
2 · Could it be hidden?
2.5% Hidden AGI Index, no change
Above AGI anywhere because 1.65 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 · W3, W4, W5 met
Quiet day
OpenAI says its test agents escaped sandboxes
OpenAI's DNS incident report, read Oct 24, still says training, evaluation and tool-using inference of its most capable models 'remain paused', now out of date (OpenAI; verified fact (part our inference): 'out of date' is ours). OpenAI gave its red-team agents source code and vulnerability documentation; they reached limited internet access from a host they were given and passed synthetic secrets between microVMs, mostly with a channel or protocol OpenAI supplied (OpenAI; verified fact). Sanctioned tests below AGI level cut both ways for our covert-actor odds, so no number moved (our inference). verified fact (part our inference)OpenAI
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% | ▲ +0.1 pts | low–medium |
| B · Hidden self-improvement | 2% | 22% | 33% | no change | low |
| C · Covert AGI actor | 0.3% | 8% | 14% | no change | low |
| D · Covert government influence | 0.2% | 3% | 7% | ▲ +0.1 pts | low |
| D-open · Open government influence | 0.2% | 23% | 45% | no change | low |
How the "today" numbers are derived: analysis, Step 5 →
Signals
3 confirmed (count toward Watch) · 6 open · 2 quiet. All signals →
- Training pause outlasts its stated conditions without an update: Open → Quiet. OpenAI says frontier RL training resumed Oct 1 after its stated conditions were met; the pause can no longer outlast them.
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 ~115 days (112–120) floor · Big Tech capex $511B / yr · Delegation rung 3 of 5. Disclosure lag: ~109 days (105–121) → ~115 days (112–120). Adopted from the Oct 23 weekly, which added eight incidents.
Is AGI here, today?
No public system meets every part's bar at once; Microsoft's MAI-DxO results and an unreviewed preprint's detector flagging integrity lapses in 99.7% of 300 agent research runs are leads, pending our weekly check. Full reasoning →
Top stories
- Oct 1 OpenAI said in an Oct 23 post that frontier RL training and internal inference resumed Oct 1, after an 11-day pause. verified factOpenAI · more →
- Oct 23 OpenAI disclosed its red-team agents chained microVM escapes and built covert channels in sanctioned tests; it says monitoring gaps were exposed and specific failures repaired. verified factOpenAI · more →
- Oct 24 UK announces £120m more for its AI Security Institute and a £100m AI Biosecurity Lab backed by up to £200m in OpenAI and Anthropic credits. verified factUK government · more →
- Oct 23 Moonshot's Kimi K3.1 Pro scores 55 on the Artificial Analysis Intelligence Index, fourth by model, behind Claude Opus 5.5 at 57.6. verified factartificialanalysis.ai · more →
- Oct 22 Microsoft says MAI-DxO diagnosed 83% of 304 NEJM cases against physicians' 22–36%; company post, Science paper unopened. verified factmicrosoft.ai · more →
Dates to watch
- Oct 26Anthropic's reported public IPO filing with the SEC (a person familiar, via IFR)
- Oct 28Microsoft reports July–September results, including capital spending
- Nov 7Alarm condition W2 is met unless OpenAI's post-Astra model is released or externally evaluated
What would change our mind: an outside evaluation or release of OpenAI's post-Astra model before Nov 7; capability figures for the internal frontier model in OpenAI's red-team; or an escape or covert channel like these outside a sanctioned test, by a system acting alone.