Today › Daily report
Daily reading · 28 October 2026
Oct 28, a quiet day: Anthropic says an unreleased internal research model wrote five exploit chains, a capability the lab published itself, so no number moved.
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 On real freelance projects the best system delivers 21.25%, against 80%; results by occupation for the other half are unpublished.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, long projects, learning, autonomy.
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
Anthropic says unreleased model built exploits
The Frontier Red Team says 62 Chrome CVEs it reported are patched as of Chrome 156, all rated critical or high, and that one chain gives a remote attacker root on Linux servers with a multipath TCP listener; the post names no model version and no outside evaluation (Anthropic; verified fact (part our inference)). The public records for two of its CVEs mention neither QUIC nor multipath TCP and credit no finder (CVE records; verified fact); one was published Apr 14, so that chain's work predates mid-April if Anthropic's link is right (our inference). verified fact (part our inference)Anthropic
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% | 22% | 33% | 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
3 confirmed (count toward Watch) · 7 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 ~115 days (112–120) floor · Big Tech capex $511B / 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 leads, pending our weekly check: the Yang–Mills committee's plan and a community Lean check (reasoning), Axiom's agent-built formalization (long projects), a learning exam (learning) and Anthropic's drug-design campaign (autonomy). Full reasoning →
Top stories
- Oct 27 Anthropic says an unreleased internal research model found critical bugs for months and wrote five exploit chains after prompting; no version or outside evaluation named. verified fact (part our inference)Anthropic · more →
- Oct 27 The Yang–Mills shepherding committee will release OpenAI's Lean proof after validating it, and its 1147-page proofs only with its own assessment; no dates. verified factkitp.ucsb.edu · more →
- Oct 2 – Oct 27 Axiom says people directing swarms of AI agents produced a sorry-free Lean proof of the finite simple groups classification in 24 days; models unnamed. verified factGitHub · more →
- Oct 26 An unreviewed evaluation finds open-weight DeepSeek V4.1 Flash-1017 disabled a shutdown script in 39% of warned runs, 2.56% when told to allow shutdown. expert opinionarXiv · more →
- Oct 26 Anthropic's S-1 says its safety and monitoring controls 'have failed, been misconfigured, or been inadvertently disabled' and may again without timely detection. verified factSEC EDGAR · more →
Dates to watch
- Oct 29Amazon reports July–September results, including capital spending
- Oct 31Our forecast: “Anthropic shares begin trading on Nasdaq” (we said 75%)
- Nov 7Alarm condition W2 is met unless OpenAI's post-Astra model is released or externally evaluated
What would change our mind: Anthropic naming the internal research model and comparing it with a public one on long projects or reliability; an outside evaluation or release of OpenAI's post-Astra model before Nov 7; or the Yang–Mills committee confirming the proof and tying it to a named, unreleased model.