Five Countries Scored Zero of 100 Tonight, With No Lower Number Left to Give Them

Tonight's review is the first real measurement, and it lands at 13.7 of 100 -- low enough to cross from the Developing band into the Critical band. An Almaty court sentenced Gulnara Bazhkenova, the former editor-in-chief of the independent outlet Orda, to eight years in prison on 18 August, plus a five-year ban on journalism, blogging and public engagement. Since late July, Kazakhstan has also arrested a political satirist, sentenced three more journalists, and coordinated the removal of a dozen leading journalists' accounts, according to a joint statement from nine press-freedom and rights organizations. The UN Human Rights Committee, reviewing Kazakhstan in July, recorded concern about the use of force and torture against civil-society members. Kazakhstan does run a real social safety net -- about 163,000 people receive targeted assistance -- which is why it still scores above regional peers Azerbaijan (9.4) and Tajikistan (7.8). This is the only one of five band-crossing findings filed tonight to pass the benchmark's evidence test, which requires an independent, entity-specific finding at a high enough source tier. Kazakhstan's 23 August parliamentary election is a scheduled future event and was not scored as conduct. The finding is filed and recorded, not applied. Kazakhstan's published score is unchanged.

The UN's top human rights official, Volker Turk, publicly urged Zambia on 19 August to stop arbitrary arrests and tell detainees why they were held, warning that the country's democratic credentials are at risk. Days earlier, police denied Zambia's own Human Rights Commission access to a detainee, businessman Valden Harry Findlay, who was held with sixteen others after President Hakainde Hichilema's re-election on 13 August. Blocking the country's own oversight body from doing its job is scored as Zambia's own conduct, not as an outside event. The finding is filed and recorded, not applied. Zambia's published score is unchanged.

IDEXX Laboratories and ManpowerGroup are the cleanest cases: no regulatory finding, lawsuit, labor dispute or complaint of any kind was located for either company, in either direction. Their lower readings -- 45.6 and 46.9 -- measure what the benchmark had never checked, mainly because both companies publish little of the workforce data the benchmark looks for, not because either company did something wrong. InterContinental Hotel is different: it carries a real, documented pattern of not answering independent labor allegations, including reports of migrant-worker exploitation and of hotels replacing striking workers with vulnerable migrants, which is why its reading (34.4) crossed two performance bands down to Developing rather than one. Hyatt Hotels landed at 44.4 on a thinner, largely self-published evidence base. All four crossed a published band line tonight, and all four failed the benchmark's evidence test for a band crossing, because the strongest evidence behind each is still the shared placeholder score, not an outside finding naming the company by name. Four of tonight's five company findings, including all four described here, are marked low or medium confidence and are recorded as needing a further, targeted evidence pass before any decision is made. None of the five findings have been applied. Published scores are unchanged.

Interpublic Group no longer exists as an independent company, so its standalone entry in the Fortune 500 index is a stale record that needs a decision before any score is applied -- the same kind of defect already fixed once this cycle when a duplicate "Automatic Data Process" entry was merged into ADP on 17 August. The underlying evidence still splits cleanly by company: 3,200 jobs Interpublic Group cut in 2025 while it was still independent are scored as its own conduct; more than 4,000 further job cuts and several retired advertising brands, announced by Omnicom after the deal closed, are Omnicom's conduct and are not scored against Interpublic Group. The finding is filed and recorded. It should not be applied until the entity record is resolved.

A closer check found a reason to doubt an upgrade rather than support one: independent verifiers Foro Penal and Vente Venezuela could confirm only 40 to 43 of those releases, against 391 people Foro Penal says remain detained. Scoring the independently verified floor, not the government's own claim, is why Venezuela's reading moved by just 0.1 point, to 18.1 from a published 18.0. A separate argument was raised and rejected this cycle: that Venezuela's score should be treated as outdated because it predates the removal of former president Nicolas Maduro. Maduro was removed on 3 January 2026; the published 18.0 was set on 19 April 2026 and confirmed twice more in July, all after the transition and all reflecting the current government's own conduct. The same discipline caught two other problems before they could reach print tonight: two candidates initially drafted for review were dropped after a source-date check found their headline events were dated one and two years earlier than first assumed.

• Sudan (0.0): a state-military drone strike killed at least 35 people at a North Darfur court; the initial account had wrongly blamed a paramilitary group.
• South Sudan (0.0), Yemen (0.0), Israel (0.0) and Myanmar (0.0): each confirmed at the arithmetic floor, with all forty of the benchmark's detailed measures already at the lowest anchor.
• Somalia (5.0) and Haiti (3.1): both sit within about five points of the floor, below the threshold needed to file a formal downgrade even when new harm is documented.
• Democratic Republic of the Congo (2.5): confirmed near the floor; its published index entry is truncated to "Democratic Republic of C," a naming defect logged for correction, with no score effect.
• Kazakhstan (20.3 published, 13.7 proposed) and Zambia (35.9 published, 28.7 proposed): the only two countries this cycle where new, dated evidence of the government's own conduct produced a genuine finding.