The most significant editorial findings in the Jul 9 briefing.
Editorial insight
A United Nations inquiry concluded on July 9 that mass killings and gang rapes committed by Sudan's Rapid Support Forces (RSF) meet the legal definition of genocide. Sudan already scores 0 of 100, the lowest possible mark, so the finding cannot push the score down further -- but it is the most serious atrocity finding this benchmark has recorded this year.
Today's question
·
boundary-watchai-lab-governancetransparency
“Anthropic sits 0.9 points below the Established boundary after a hidden Claude Code tracker was confirmed and removed; would a second undisclosed telemetry episode, or a regulator finding of deceptive practice, move the Integrity dimension enough to cross that boundary before the EU AI Act's August 2 enforcement date?”
Sudan holds at 0 of 100, the lowest possible score. A United Nations inquiry concluded on July 9 that mass killings and gang rapes committed by the Rapid Support Forces (RSF), the paramilitary group fighting Sudan's army, meet the legal definition of genocide.
Why it matters
Two governments accused of the world's worst atrocities already hold the lowest possible score, so today's grim milestones cannot push either score down any further.
Temporary Protected Status work authorization expires for roughly 350,000 Haitians in the United States. This is United States conduct, not scored against Haiti.
United States-Iran talks resume in Doha following the close of the state funeral window. A verifiable agreement would be a positive scored event; a breakdown combined with new provocation would be …
The natural seasonal peak for the current Ebola outbreak closes. Documented government obstruction of WHO response corridors is the trigger toward the absolute floor.
The EU AI Act becomes fully applicable. A second undisclosed telemetry episode or a regulator finding of deceptive practice would be the next event that could move the score.
The 2027 election under the July 2025 indefinite re-election amendment. Restoring due process and reinstating civil society are the conditions that would move the score back up.
First confirmed enforcement action under the new Ethnic Unity Law's diaspora-liability clause -- a prosecution, forced return, or family detention abroad.
Independent confirmation of the forensic pathologist's undercount allegation, or documented aid obstruction, is the next event that would move the score toward a formal change from 18.0 of 100.
How to read this briefing— Bands, scores, and terms
Schema guide
The 5 performance bands
critical0–20
developing20–40
functional40–60
established60–80
exemplary80–100
Two scales
Each of the 8 dimensions is scored 1.0–5.0; these combine into a 0–100 composite score, mapped to the 5 bands above.
Key terms
Band crossing
A score change large enough to move an entity from one performance band into an adjacent one — the most structurally significant finding in a given cycle.
Boundary watch
An entity whose current score is within 3 points of a band threshold, flagged for priority reassessment in the next cycle.
Carry-forward
A dimensional credit retained from a prior assessment when new evidence is insufficient to revise a specific dimension; disclosed explicitly.
First baseline
An entity's inaugural composite score — no published score exists to compare against, so delta is not shown.
Floor designation
The most serious finding: all 8 dimensions resolve at the lowest behavioral anchor (1.0/5.0) across multiple cycles, yielding a composite of 0.
Forward trigger
A scheduled future reassessment event (e.g., a policy implementation date or legislative deadline) that may materially change an entity's score.
One AI company got caught quietly tracking some of its own users; the other was mistakenly described as scoring far higher than it actually does. Neither changes the published record tonight.
Anthropic holds at 59.1 of 100, in the Functional band, 0.9 points below the higher Established band.
Where this sits
Read the full signal
An independent researcher discovered that Anthropic's Claude Code product had, since roughly March 2026, secretly flagged certain users -- those whose device location matched parts of China -- by hiding a tracking signal inside normal-looking text. An Anthropic engineer confirmed the code was real and described it as an unannounced experiment meant to catch rival AI labs using Claude's output to train their own models without permission. Anthropic removed the tracker around July 1 once the discovery became public. Undisclosed user tracking is a genuine honesty problem, especially for a company that has publicly opposed government surveillance demands. But because this was a single episode that Anthropic acknowledged and fixed once caught, it does not add up to enough to move the score. Meta Platforms holds at 7.8 of 100, in the Critical band -- close to the very bottom of the Fortune 500 ranking, not 60 of 100 as an earlier internal note mistakenly described it. Court filings this week show four US states are seeking about $1.4 trillion in penalties, alleging Meta designed Facebook and Instagram to be addictive to young users. Legal experts say that figure is a maximum possible demand, not a likely outcome, and the underlying case was already the basis for Meta's low score. Confirmed at 59.1 of 100 for Anthropic and 7.8 of 100 for Meta; confidence: high for both.
Israel keeps approving new West Bank settlements and Yemen's Houthi authorities keep holding UN staff in detention -- both patterns that already put each score at the floor.
Israel holds at 0 of 100. Its cabinet approved 13 new settlements in the central West Bank on July 6, a move Finance Minister Bezalel Smotrich called a revolution in settlement policy.
Where this sits
Read the full signal
Settlers have also destroyed an electricity line, vandalized greenhouses, and attempted to seize livestock from Bedouin herders in the same window, continuing a pattern of violence UN monitors say has surged since 2023. Yemen holds at 0 of 100. The Houthi movement, which controls much of the country, continues to detain 73 United Nations and civil-society staff, some held since 2024; a World Food Programme worker previously died in Houthi custody. Both patterns are each government's own conduct, not conduct by an outside party, and both already sit at the lowest possible score, so neither can move lower. Confirmed at 0 of 100 for both; confidence: high for both.
Congo's Ebola outbreak is now the fastest-growing on record, and Uganda reported a completely separate, related virus outbreak the same week.
The Democratic Republic of the Congo holds at 2.3 of 100, just above the absolute floor. Its Ebola outbreak reached 600 confirmed deaths and 1,759 confirmed cases as of July 9 -- described by health authorities as the fastest-growing Ebola outbreak ever recorded.
Where this sits
Read the full signal
Only 82% of the more than 10,000 people who had contact with confirmed cases are being tracked, short of the 95% the World Health Organization says is needed to stop the spread. Ebola is caused by a virus, not by government action, so the outbreak itself is not scored against the Congo's government -- but the tracking gap is a real, if not new, sign of weak response capacity, consistent with the country's already-low score. Uganda holds at 20.3 of 100, in the Developing band, just 0.3 points above the Critical band. Uganda reported a completely separate outbreak of Marburg virus, a relative of Ebola, on June 30, even as its own Ebola cluster has shown no new case in more than two weeks. Uganda's prompt report to the World Health Organization is treated as a small positive sign about its own response, not a mark against it. Confirmed at 2.3 of 100 for the Democratic Republic of the Congo and 20.3 of 100 for Uganda; confidence: high for both.
Venezuela's earthquake-response score keeps falling just short of the point threshold that would trigger a public score change, for the third night in a row.
Venezuela holds at 18 of 100. Its earthquake death toll rose to 3,899 as of day 16 (July 9), up from 3,535 two days earlier, now the deadliest natural disaster in the country in more than a century; US Geological Survey modeling suggests the true final toll could exceed 10,000.
Where this sits
Read the full signal
Recalculating Venezuela's disaster-response performance for this evidence would put the score at 15.3, a decline of 2.7 points -- the same reduced reading confirmed on July 7 and July 8, still short of the 5-point threshold the benchmark requires before changing a published score. Four other countries hold at their published scores tonight while carrying a documented downward-trending watch: Nigeria (18 of 100), where 17 million or more people face crisis-level hunger across the north; China (19.5 of 100), where a new ethnic-assimilation law continues to be enforced; Pakistan (17.2 of 100), where more bodies were recovered from a Balochistan region enforced-disappearance pattern showing signs of torture; and Lebanon (17.2 of 100), where the main source of harm remains outside fighting rather than the Lebanese government's own conduct. None of these five crosses the threshold for a score change tonight. Confirmed at 18 of 100 for Venezuela and Nigeria, 19.5 of 100 for China, and 17.2 of 100 for Pakistan and Lebanon; confidence: high for all five.
Sudan and Myanmar -- Gravest Atrocity Findings of the Series, Both Floor-Confirmed
A UN inquiry finds Sudan's RSF conduct meets the legal test for genocide while Myanmar's civil war passes 100,000 deaths; both countries hold at 0 of 100.
Read the full signal
Sudan (0 of 100): the United Nations concluded July 9 that mass killings and gang rapes by the Rapid Support Forces meet the legal definition of genocide, the most severe atrocity finding recorded this year.
Myanmar (0 of 100): the post-coup civil war has now killed more than 100,000 people, with the military responsible for about 7 in 10 civilian deaths.
Both countries already hold the lowest possible score, so neither finding can push the score down further -- it is recorded as confirming, not lowering, the existing score.
·medium
AI Labs and Fortune 500 -- Anthropic and Meta Both Hold Steady After Special Review
Anthropic's hidden Claude Code tracker and Meta's corrected 7.8-of-100 score both confirm without a change tonight.
Read the full signal
Anthropic (59.1 of 100): a hidden tracking feature in Claude Code, active since roughly March 2026, was discovered and removed; an Anthropic engineer confirmed it was an undisclosed experiment to detect misuse by rival AI labs.
Meta Platforms (7.8 of 100): near the bottom of the Fortune 500 list, not 60 of 100 as an earlier internal note mistakenly described it. A $1.4 trillion penalty demand from four US states is a litigation maximum figure, already reflected in the score.
Both hold at their published scores tonight with a documented downward-trending watch carried forward.
10 signals shown
Risk signals
Developments that may affect future scores. Watch items from the Jul 9 briefing.
Risk
Sudan's genocide finding and Myanmar's 100,000-death milestone both document worsening severity that the lowest possible score cannot reflect with a further point drop.
Risk
Anthropic sits 0.9 points below the higher Established band after a hidden tracking feature in Claude Code was discovered and removed.
Risk
The Mali-Burkina Faso scoring question has now run 16 consecutive days without a decision from the benchmark's coordinating team.
Risk
Venezuela's earthquake-response score has held at a 2.7-point sub-threshold decline for three consecutive checks with no independent confirmation of the undercount allegation.
Risk
Central Africa now has two active hemorrhagic-fever outbreaks at once: Congo's Ebola outbreak past 600 deaths and Uganda's separate Marburg outbreak.
Risk
Temporary Protected Status work authorization for roughly 350,000 Haitians in the United States expires July 10.
Risk
Eleven of the 15 countries and companies assessed tonight carry a documented downward-trending watch -- the highest share recorded so far in July.
Score movements
All entities assessed this cycle. No score changes.
Next signal: — The EU AI Act becomes fully applicable. A second undisclosed telemetry episode or a regulator finding of deceptive practice would be the next event to watch.
Next signal: — A coordinator-level review comparing Mali's score to Burkina Faso's remains open, now treated as a standing item for the benchmark's coordinating team.
Next signal: — US Temporary Protected Status work authorization expires for roughly 350,000 Haitians. This is United States conduct, not scored against Haiti.
A documented own-conduct deterioration in epidemic response, such as government obstruction of WHO response corridors, would be the next event that could push Uganda back into the Critical band.
A confirmed enforcement action under the new Ethnic Unity Law's diaspora-liability clause -- a prosecution, forced return, or family detention abroad -- is the next threshold event to watch.
Documented expulsions of Bengali Muslim and Rohingya populations, and continued digital surveillance targeting minority groups, remain under watch. Not assessed tonight.
A second undisclosed telemetry episode, evidence the removed tracker was broader or retained longer than stated, or a regulator or court finding of deceptive practice would each independently move the score toward a change. The EU AI Act becomes fully applicable August 2.
Independent confirmation of the forensic pathologist's undercount allegation, or documented aid obstruction, is the next event that would move Venezuela's score down.
Documented government obstruction of World Health Organization response corridors would be the trigger to move toward the absolute floor. The natural seasonal peak for this outbreak runs through July 31.
A durable ceasefire combined with the Lebanese Armed Forces taking sovereign control of southern Lebanon and delivering reconstruction would together move Lebanon toward 20 of 100.
Independently verified deliberate targeting of civilians in a cross-border strike would move Pakistan's score down; durable relief for detained protesters and disappeared Balochistan residents would move it up.
A formal famine declaration covering Buur Hakaba or Baidoa, combined with documented government aid obstruction, is the next scored trigger. The famine-risk window runs through September 2026.
Documented atrocities by Haitian state security forces, or deliberate relief obstruction, would move the score down. Verified state-led civilian protection would move it up.
The outcome of the Justice Department's False Claims Act complaint or the Barrows v. Humana class action over AI-driven claim denials would confirm the current Developing-band placement either way.
Due-process restoration, civil-society reinstatement, and reversal of the indefinite re-election amendment remain the path back toward 20 of 100. The 2027 election under the amendment is the next scheduled test.
Documented discriminatory denial of emergency protection to stateless bidoon or migrant workers during wartime would move Kuwait's score down; durable expansion of bidoon rights would move it up.
Independently documented deliberate targeting of Palestinian civilians by Palestinian authorities themselves, or systematic diversion of aid, would move the score down.
A coordinator-level review comparing Mali's conduct against Burkina Faso's (6.3 of 100) is the next scored event needed to resolve this gap. No review date has been set.
A merits ruling or verdict for the states, a material settlement, or new dated evidence of youth-safety harm or internal-findings suppression is the next event that would move Meta's score.
A fourth nationwide blackout, or a documented healthcare-access failure beyond tens of thousands of canceled surgeries, is the next event that would move Cuba's score toward a formal change.
A return to lethal crackdown tactics, such as the 41 protester deaths recorded during the 2025 Saba Saba march, would be the next event that moves Kenya's score down toward the Critical band.
documented
Evidence ledger
Primary sources reviewed in this briefing cycle. 11 sources linked.
The Irrawaddy, July 2026: Myanmar's post-coup civil war death toll surpassed 100,000, with the military responsible for the majority of civilian fatalities.
Malwarebytes, July 2026: Anthropic confirmed that a hidden tracking feature in Claude Code, active since around March 2026, flagged users in certain geographic locations and was removed after discovery, with the company describing it as an unannounced experiment.
ConsumerAffairs, July 8, 2026: four US states are seeking roughly $1.4 trillion in penalties from Meta, alleging Facebook and Instagram were deliberately designed to addict young users.
Al Jazeera, July 6, 2026: Israel's cabinet approved 13 new West Bank settlements, which Finance Minister Bezalel Smotrich described as a revolution in settlement policy.
Al Jazeera, July 9, 2026: confirmed Ebola deaths in the Democratic Republic of the Congo reached 600, in what health officials describe as the fastest-growing outbreak on record.
Daily Post, July 9, 2026: the World Food Programme warns that food assistance cuts are imminent across northern Nigeria as hunger deepens faster than projected.
Decrypt, July 2026: Anthropic removed the hidden Claude Code tracker after backlash over its undisclosed use.
Floor designations
·8 entities at composite 0 with documented evidence pattern
Composite scores resolving at zero — methodology disclosure
These entities consistently score the worst result across all 8 dimensions of compassionate conduct — the benchmark's most serious classification.
What “floor” means: every one of the 8 dimensions (Recognition, Response, Reduction, and 5 others) resolves at the lowest behavioral anchor (1.0/5.0) across multiple assessment cycles, yielding a composite score of 0. Full methodology.
Daily briefings surface the headline finding. Full benchmark reports include all 40 subdimension scores, complete evidence trails, certified assessments, and sector-level analysis packages — the record researchers and journalists cite.
Independence note: entities never pay for inclusion, score changes, or suppression of findings. Commercial services support access, interpretation, and institutional use only.