Nigeria 21.9 → 18.0 (Δ -3.9): IPC Phase 5 Catastrophe for ~15,000 Borno Residents — Band Crossing Developing → Critical on Sub-Threshold Delta

Cal.) alleging that Figure AI operated humanoid robots at force levels more than twice the threshold necessary to fracture an adult human skull, with no formal safety procedures, incident-reporting systems, or risk-assessment processes in place. A malfunctioning robot allegedly struck a refrigerator with a quarter-inch-deep gash in the stainless-steel door while narrowly missing an employee. An investor-reviewed home-deployment safety roadmap was allegedly downgraded after the company's ~$39B valuation round — flagged internally as potentially fraudulent — with an e-stop initiative and a safety feature both abandoned. Gruendel was allegedly terminated after escalating these concerns. Figure AI denies the allegations and states Gruendel was terminated for poor performance. The allegations are pre-adjudication and reported by multiple outlets; they are scored as credible, specific Tier-2/3 evidence of an absent internal safety governance function and retaliation against the designated safety voice. Dimension movements: AWR 2.5 → 2.0, BND 2.5 → 2.0, ACC 2.5 → 2.0, INT 2.5 → 2.0. Figure AI remains in the Developing band at 31.3. Confidence is medium given pre-adjudication status.

ICC Prosecutor Karim Khan characterized RSF atrocities as being 'repeated town by town,' indicating documentation of a geographic atrocity pattern. Sudan's absolute harm-flag floor at 0.0 absorbs the evidence without composite movement. Al-Obeid represents the most precisely defined imminent large-scale atrocity risk in the benchmark's current monitoring portfolio, concentrated in a single accessible population center rather than dispersed across conflict lines.

The investigation scope covers advertising practices, user engagement and retention design, model sycophancy, consumer and health data handling, and protection of minors and seniors. This is the broadest multi-jurisdictional consumer-protection action yet opened against a major AI lab. The assessed delta of -4.7 lands below the 5-point proposal threshold with no band change; the investigation is an opened probe, not adjudicated harm. OpenAI confirms at 27.5 Developing. Re-flag when charges are filed or findings published.

Contact tracing remains below 20% of estimated exposures. No Bundibugyo-strain-specific vaccine exists. The Mongbwalu burial-team hostage event from the prior cycle (5 workers) represents a direct threat to response capacity. The July natural Ebola peak window has narrowed to approximately 1-5 weeks. DRC confirms at 2.3 Critical under the floor-limitation methodology; state obstruction of the WHO-led response is the conversion trigger for a floor-designation proposal.

The scanner re-surfaced the Senate PSI and HHS-OIG denial-rate evidence that drove the June 18 downgrade; no genuinely new June-2026 Humana-specific federal action was confirmed beyond what is already priced in. Humana sits 0.6 points above the Functional-to-Developing boundary at 40.0; CMS enforcement action on Medicare Advantage prior-authorization practices remains the conversion trigger for a Developing-band proposal.

• The composite delta (-3.9) is below the standard 5-point change threshold; the band-crossing rule applies independently of delta magnitude when the proposed composite crosses a designation boundary.
• Nigeria is the anchor case for distinguishing 'prevention failure' from 'active perpetration': the harm is Boko Haram/ISWAP conflict-driven; the scoring failure is the state's failure to prevent, protect, and prioritize response to foreseeable catastrophic need.
• The sub-threshold-delta band crossing opens a methodological category not previously applied in the June 2026 series — and a candidate for v1.3 formalization as the 'structural-deterioration band crossing' anchor case.
• The prior June 2026 band crossings (US -5.9, Bolivia -10.0, UAE -5.0, India -7.1) all had above-threshold composite deltas; Nigeria is the first exception and tests the band-crossing rule's independence from the proposal threshold.