A Bangladesh Court Ruling Pulls the Benchmark in Two Directions at Once

The mission found security forces killed people in all 31 of Iran's provinces, that at least 221 children were killed, some as young as two, and that at least 29 men were executed between March and August 2026 after expedited trials, with more than 70 people — including three boys — still at risk of execution. Iran already sits near the benchmark's absolute floor, at 2.5 out of 100, the third-lowest of 191 countries measured. That number does not move tonight. What changes is the strength of the evidence behind it: the case for 2.5 previously rested mainly on watchdog tallies, and now carries a United Nations mandate's own finding of crimes against humanity. A move all the way to the benchmark's zero floor was considered and set aside for now — the benchmark's own rule requires two independent high-tier sources before a floor designation, and this cycle located one.

There is no annual report, no workforce data, no safety record and no sustainability disclosure to read for any period after 2020. Key Energy Services is nonetheless published in the Fortune 500 index at position 500, carrying a score of 21.9 out of 100. Assessing 40 separate behaviors for a company with no current public record would measure its silence rather than its conduct, so no score was issued this cycle. The finding is about the benchmark's own catalogue, not about the company: the same class of problem the benchmark identified in August 2026, when it found Halodi Robotics and 1X Technologies were the same company listed twice. Until a decision is made about whether deregistered, privately held companies belong in the index at all, this entity is being set aside from further nightly review rather than resurfacing each cycle as an unscorable candidate.

A routine rotation check this cycle found that this score is not an old placeholder — it traces to a full assessment on file from April 22, 2026, filed under a mangled filename that had made it look, incorrectly, like Macy's had never been individually researched. Re-reading the same public record this cycle produced 37.5 out of 100, which would cross the line into Developing. That reading was not filed. The entire gap amounts to 0.15 points on a five-point scale — well within the range one researcher's reading can differ from another's on the same facts — and this review found zero new evidence about Macy's conduct in the two weeks it covers. Macy's did cut roughly 1,050 jobs and close 14 stores in 2026, filing the required 60-day notices in January; it also pays for a fully funded education benefit for every employee and negotiated severance of up to 52 weeks for unionized staff in California. None of that is new. Macy's published score stays at 41.3, and a single confirmed new finding about Macy's labor or customer conduct would be enough to make the tier question testable on evidence rather than re-reading.

No individual assessment of the insurance and risk-advisory firm exists on file, despite its position at 42nd of 447 companies in the Fortune 500 index with a published composite of 60.9. A short first pass — four searches, not a full 40-point review — produced an indicative reading 24.6 points lower, which would cross two tier lines. That reading was not filed and should not be read as a measurement: independent raters disagree with it (MSCI rates the firm AA, a leader grade, and Sustainalytics puts its risk at 19.5, both low-risk readings), and the company's own sustainability report was not read this cycle. What the pass did establish on the record: $187 million in restructuring costs, mostly severance, under a three-year program called Thrive, and a health-plan lawsuit the firm must now answer rather than having had dismissed. Marsh & McLennan is referred for a dedicated full assessment rather than a routine nightly check, the same treatment the benchmark gave a cluster of placeholder robotics-industry scores in August 2026.

Because the harm was inflicted by a foreign state at war with Ukraine, it produces no change to Ukraine's own score — the benchmark measures an institution's own conduct, and scoring a country down for being attacked would penalize the wrong actor. What is scoreable is Ukraine's own response. Its State Emergency Service reported casualties through official channels within hours across four regions, and rescue crews reached trapped residents — a functioning alert-and-rescue chain. Physical shelter is thinner: as of September 3, five days before the strike, Kyiv had only one certified public modular shelter, funded by a private company rather than the state, after a summer of strikes that killed around 150 civilians in the city. A law requiring shelters in new construction took effect on September 8 — the day of the strike itself. Those facts fall just outside this review's evidence window and are logged as watch items rather than scored. Separately, and not examined this cycle, Ukraine's own human rights commissioner has called draft-enforcement abuses at recruitment centers "systematic and large-scale" — the most likely source of a future downward reading.

Five proposed federal class actions followed within days, filed by customers in Texas, Indiana and Minnesota; those remain allegations, untested in court. A widely circulated figure claiming 7.49 million records were affected was not used anywhere in this assessment: it originates from the hacker's own claim, appears on an undated website, and CenterPoint has not confirmed it. This is CenterPoint's first individually researched score, produced without prior history to compare against. It measured 33.8 out of 100, 2.1 points below the published 35.9 — inside the benchmark's normal confirmation range. The utility's income-based Gas Affordability Program, which credits charges above 6% of a household's annual income, is the strongest thing in the record; the weakest is that its own problems — the 2024 Hurricane Beryl outage failures, this breach, and a billing practice that applied payments to unregulated charges before regulated gas bills — have each surfaced through outsiders rather than the company's own monitoring.