Threat+Constellation
Live— UTC

Attack Efficiency Analysis

Observed Confirmed Success Rate = confirmed successful attempts / eligible observed attempts, only inside one bounded population.

LAC disclosure count is not an efficiency rate

LAC's 119 public disclosures (through 2026-10-06) are a public_incidents series. Dividing them by unmatched attack attempts returns NOT_ESTIMABLE.

Numerator and denominator are not the same bounded population or unit.

Unverified success

INSUFFICIENT_EVIDENCE: Unverified success cannot be counted as confirmed compromise.

Scenario A — high volume, low success
Attempts8000Confirmed180
  • Attempts: 8000 · OBSERVED
  • Confirmed: 180 · OBSERVED

Scenario A — high volume, low success

Unit: synthetic attack_attempts

Time window: synthetic-t0 → synthetic-t1

Measurement scope: Scenario A bounded cohort

Data origin: SYNTHETIC

SYNTHETIC — NOT REAL-WORLD OBSERVATION

Scenario B — low volume, higher success
Attempts340Confirmed133
  • Attempts: 340 · OBSERVED
  • Confirmed: 133 · OBSERVED

Scenario B — low volume, higher success

Unit: synthetic attack_attempts

Time window: synthetic-t0 → synthetic-t1

Measurement scope: Scenario B bounded cohort

Data origin: SYNTHETIC

SYNTHETIC — NOT REAL-WORLD OBSERVATION

Cohort comparison is descriptive. A higher synthetic success rate does not explain a real-world change in attacker skill, coverage, or disclosure.

Evidence Confidence

  • PUBLIC_SOURCE · STATED · VERIFIED

    国内で相次ぐWebシステムからの情報漏えい～ラックの観測から見える変化と、企業が今とるべき対策

    Empirical: yes, scoped

  • PUBLIC_SOURCE · STATED · VERIFIED

    Cyber Emergency Center dispatch mix, server intrusion share

    Empirical: yes, scoped

  • PUBLIC_SOURCE · STATED · VERIFIED

    Days from detection/recognition to first public disclosure

    Empirical: yes, scoped

  • PUBLIC_SOURCE · INFERENCE · VERIFIED

    Hypothesis: fewer attempts may still produce more efficient compromise

    Empirical: no

  • PUBLIC_SOURCE · EXCLUDE · VERIFIED

    No public evidence of AI-agent involvement in the Japanese series

    Empirical: no

  • PUBLIC_SOURCE · EXCLUDE · UNVERIFIED

    Overseas AI-harness campaign reported by Gambit Security

    Empirical: no

  • SYNTHETIC · STATED · VERIFIED

    AEIE synthetic scenario lab

    Empirical: no

Evidence confidence panel

Unit: record / classification

Time window: access date 2026-10-21

Measurement scope: AEIE evidence registry

Data origin: mixed; synthetic records cannot become empirical

EMPIRICAL / SOURCE-BACKED