Daily intelligence on AI, institutions, power systems, and measurable human impact.
We translate global events into compassion risk signals, institutional accountability insights, and early warnings for systems that shape human lives. Every finding is grounded in primary-source evidence — litigation records, regulatory filings, investigative reporting, and international legal instruments.
Research agents scan all 1,155 benchmarked entities nightly for new evidence. Flagged entities receive full 40-subdimension assessments. Score changes are proposals, not automatic updates — a human analyst reviews each before published scores change.
The most material cross-entity intelligence from tonight's research. Each signal is grounded in primary-source evidence and sorted by severity.
Critical severityMethodology·Signal 01·
Floor-Limitation Cluster — CRITICAL Methodology Gap
What happened
Five entities (xAI 2.2, Palantir 6.6, Sudan 0, South Sudan 0, Israel 8.8) accumulating material new evidence each night with zero expressive composite movement under current methodology. Gap is now 7 nights overdue for a methodology response.
Why it matters
New in-window evidence this cycle: Sudan al-Fasher mass detention April 27 (cholera, ethnic-targeted torture); Israel OCHA April 23; xAI French criminal summons; Palantir IRS expansion. The published scores cannot differentiate between these entities on severity of harm. Recommend dedicated methodology development cycle within 7 days.
47,000+ tech job cuts in 30 days across three major F500 entities. Oracle 30,000 (March 31, pre-window, index gap), Meta 8,000 announced April 23, Microsoft voluntary buyouts for 8,750 announced April 23.
Why it matters
All three cite AI spending reallocation as explicit rationale. This is the largest AI-driven labor displacement event in benchmark history. Sector mean EMP dimension will face downward pressure if these composites are reassessed post-confirmation.
Sudan enters fourth year of civil war (April 15 anniversary): 400,000 dead estimate, 34M in acute humanitarian need, RSF holding 1,470+ civilians in mass detention as of April 27.
Why it matters
South Sudan: UNMISS mandate expires April 30 in 2 days with active civil war, Abiemnom massacre (178 dead, March 2026), and famine risk confirmed by UN April 17. Both entities are at scoring floor (composite 0) with accumulating evidence the methodology cannot register.
DeepMind and Anthropic producing strongest positive governance signals in the pipeline since inception. DeepMind Frontier Safety Framework v3 (April 17) adds Tracked Capability Levels and Critical Capability Level for harmful manipulation — proactive safety expansion 96 days before EU AI Act enforcement.
Why it matters
Anthropic April 22 DC Circuit filing explicitly defends Claude non-manipulability and refuses mass-surveillance and autonomous-weapons use. Both confirm without proposals. Divergence from xAI (floor 2.2), Palantir (floor 6.6), and OpenAI (trial, 27.5) is widening.
·6 entities at composite 0 with documented evidence pattern
Composite scores resolving at zero — methodology disclosure
These entities have all 8 dimensions resolving at the lowest behavioral anchor (1.0/5.0) across multiple assessment cycles. The composite is zero by formula; each entity page surfaces the documented evidence pattern. Floor designation is reversible when functional improvement is evidenced. Read the methodology.
Primary-source alerts from overnight scanning. Each alert is linked to original regulatory filings, court records, investigative reports, and international legal instruments.
AI Labs — Positive Governance Divergence
Fortune 500 — AI-Labor Cascade
Floor-Limitation Cluster — CRITICAL Methodology Gap
Countries — Active Conflict Crisis
Get the full benchmark report
Daily briefings surface headline findings. Full benchmark reports include complete methodology documentation, all 40 subdimension scores, full evidence trails, certified assessments, and sector-level analysis packages.
Floor-limited confirmation. April 20 French summons of Yaccarino and Musk (CSAM hearing), Amsterdam injunction, EU Parliament deepfake ban all in prior baseline. No new acute April 25-28 evidence. Composite 2.2 holds at floor. Floor-limitation methodology gap now 7+ nights overdue for this entity.
Trial Day 2 opening arguments delivered April 28. Two surviving claims: unjust enrichment and breach of charitable trust. Musk seeks $134B redirected to OpenAI charity. Judge Gonzalez Rogers makes final remedy decision; jury advisory only. Composite 27.5 Developing appropriately reflects unresolved governance controversy. Post-trial mandatory re-queue after May 21 liability phase and remedy decision.
Floor-limited confirmation. Composite 6.6 holds (rebaselined April 27 from 10.3 calibration proposal). April 24 Intercept IRS scope expansion already absorbed. UK Parliament April 16 debate, 32-officials revolving-door pattern, 13-employee whistleblower letter all in prior captures. Floor-limitation methodology gap continues.
April 22 96-page DC Circuit appeals filing is a substantive positive INT/ACC signal — Anthropic explicitly defends Claude classified-network non-manipulability and maintains refusal to permit mass domestic surveillance and autonomous weapons use. Parallel SF federal court win removed stigmatizing labels. SB3444 opposition (vs OpenAI support) further supports INT positioning. Composite 61.6 Established holds.
April 23 8K layoffs plus 6K cancelled open roles announced with AI-efficiency rationale. NM jury ($375M potential) and LA jury ($6M with Google) verdicts confirm rather than newly add to BND/INT evidence. Sub-5pt EMP calibration available but not warranting proposal. Composite 29.4 Developing holds.
Floor-limited confirmation. Composite 0 holds at absolute floor. New in-window: Al Jazeera April 27 reports 1,470+ civilians and 907 military and 20 doctors held in RSF detention with cholera outbreak and ethnic-targeted torture and execution. UN Security Council April 28 added new Resolution 1591 sanctions designations. Genocide finding remains institutional anchor. Floor-limitation methodology gap continues.
April 23 first-ever voluntary buyout in 51-year history: approximately 7% of US workforce (8,750 employees), Rule of 70 eligibility, voluntary opt-in. Materially less coercive than Meta/Oracle direct layoffs but characterized as a buyout dressed as a benefit by independent analysts. Sub-5pt EMP calibration available; composite 66.4 Established holds. DATA INTEGRITY FLAG: user input claimed 76.8 baseline; live data shows 66.4 — reconciliation needed.
April 17 Frontier Safety Framework v3 published — adds Tracked Capability Levels, new Critical Capability Level for harmful manipulation, extends misalignment protocols to large-scale internal deployments. Positive INT/SYS signal pre-EU AI Act enforcement (96 days). Lacks specified independent external oversight which limits an ACC upgrade. Composite 65.0 Established holds. First substantive evidence-based confirmation for this entity.
Floor-limited confirmation. Composite 8.8 holds at near-floor. OCHA April 23 Humanitarian Situation Report: dire conditions continue but aid entry surged 14-20 April with Zikim Crossing reopening (partial mitigation). 37 NGO suspension and Israeli High Court injunction still in litigation. April 28 UN Security Council Gaza ceasefire briefing held. Amnesty 2026 quasi-judicial finding remains institutional anchor.
California AG Bonta April 20 unsealed antitrust evidence showing Amazon directed brands (Levi's, Hanes) and rival retailers (Walmart, Target, Best Buy, Home Depot, Chewy) to inflate prices on competing platforms. Trial 2027. Composite 17.8 Critical already deeply reflects accumulated evidence. Sub-5pt INT calibration available; not warranting proposal.
Treasury cancellation of all 31 contracts ($21M) following IRS contractor unauthorized-disclosure conviction already absorbed in April 25 baseline. No new April 25-28 evidence. Composite 36.7 Developing holds.
Composite 10.9 Critical already deeply reflects accumulated evidence. April 23 8K layoffs and child-safety AI scandal and jury verdicts confirm rather than newly add to baseline.
T3 rotation. Tier-1 LGBTQ rights globally; same-sex marriage 2013; anti-discrimination law 2004. Transgender life expectancy approximately 45 years and physical violence rates represent EQU ceiling. Composite 83 Exemplary holds.
T3 rotation. Strong civil liberties (Freedom House 2026); same-sex marriage entered force January 2026; GRECO May 2025 noted gaps re prince's prosecutorial-block authority; UN expert flagged financial-secrecy-as-rights-harm; 2026 GIR MCAA tax-cooperation. Composite 83 Exemplary holds.
T3 rotation. Nine consecutive years 100% on HRC Corporate Equality Index; Forbes Best Employers for Diversity 2018-2023; five perfect Disability Equality Index scores; pension plan offered (rare in sector); year 10 of 15-year diversity roadmap. Composite 77.9 Established holds.
Key highlights
Editorial-level findings from the Apr 28 research cycle.
01
Zero proposals tonight — highest-ever confirmation-dominant pipeline night (18/19 confirmed, 0 proposals). The clean state is real but partially reflects floor-limitation methodology boundaries rather than universal score stability.
02
South Sudan hard-deadline breached: sixth consecutive deferral with UNMISS mandate expiring April 30 (T-2 days). Human review queue escalation is now mandatory. Vote outcome cannot be expressed via composite movement due to floor-limitation — methodology note registration required.
03
Strongest positive AI governance signals since pipeline inception: DeepMind Frontier Safety Framework v3 (April 17, Tracked Capability Levels, Critical Capability Level for harmful manipulation) and Anthropic 96-page DC Circuit filing (April 22, explicit refusal of mass-surveillance and autonomous-weapons use) both confirm without proposals.
04
AI-labor sector cascade intensifies: 47,000+ tech cuts in 30 days (Oracle 30K pre-window, Meta 8K April 23, Microsoft 8,750 April 23). AI spending reallocation is the stated rationale across all three. Oracle cannot be anchored — not in live Fortune 500 index; index reconciliation required.
05
Floor-limitation methodology gap escalated to CRITICAL: 5 entities (xAI, Palantir, Sudan, South Sudan, Israel) with accumulating evidence across 7 consecutive nights. Methodology development cycle recommended within 7 days.
06
Musk v. OpenAI trial Day 2 (April 28 opening arguments) proceeds with two surviving claims. Fraud counts dropped pre-trial. Liability phase to May 21; post-remedy mandatory re-queue for OpenAI. Trial outcome is a deterministic INT/SYS inflection point.
The weekly briefing on institutional compassion scores
Score changes, sector trends, and emerging risk signals from overnight research across 1,155 entities — every Monday. Free.
No spam. Unsubscribe anytime. Your email is never shared.
Sector intelligence
Analyst-level observations on patterns emerging across indexed sectors from the Apr 28 research cycle.
AI Labs
›Internal divergence accelerating. DeepMind and Anthropic confirmed in Established band (65 and 61.6 respectively) with positive governance evidence. OpenAI (27.5 Developing) held pending trial outcome. xAI (2.2 Critical) and Palantir (6.6 Critical) floor-limited with accumulating negative evidence. Meta AI (29.4 Developing) confirmed amid layoff and child-safety evidence. EU AI Act enforcement 96 days away (August 2, 2026) is a sector-wide inflection approaching rapidly. Entities with proactive safety frameworks (DeepMind, Anthropic) are better positioned for compliance than those with ongoing regulatory conflicts.
Fortune 500 — Technology
›AI-labor displacement is the dominant sector signal. Three major F500 tech firms announced or executed large-scale workforce reductions in the April 14-28 window, all citing AI spending reallocation. Combined total exceeds 47,000 positions. Oracle layoffs (30K, March 31) cannot be scored due to index gap. Microsoft and Meta both confirmed at current composites. The pipeline should treat the AI-labor cascade as a standing T1 sector alert through Q2 2026.
Conflict Countries
›Sudan and South Sudan both at composite 0 floor with accelerating humanitarian deterioration. Sudan: fourth year of war, RSF mass detention with cholera and ethnic-targeted torture documented April 27, 400K dead estimate. South Sudan: active civil war reignited, UNMISS mandate expiry imminent, famine risk confirmed. Both entities are accumulating evidence each night that cannot be expressed via composite movement. The floor-limitation methodology gap is most consequential for these two entities.
Fortune 500 — Defense/Intelligence Contractors
›Booz Allen Hamilton confirmed at 36.7 Developing (first evidence baseline April 25). Palantir confirmed at 6.6 Critical (floor-limited). Both entities are in active accountability crises: Booz Allen via IRS contractor conviction and Treasury contract cancellation; Palantir via IRS data mining scope expansion, UK Parliament debate, 13-employee whistleblower letter, and 32-officials revolving-door pattern. The federal contractor accountability dimension is structurally weak across both entities. Oracle (index gap) is the third major federal contractor with unprocessable new evidence this cycle.
Emerging risks
Forward-looking risk signals from the Apr 28 research cycle. These are not current findings — they are early warning flags.
Risk
Musk v. OpenAI trial (liability phase to May 21): breach-of-charitable-trust claim is potentially precedent-setting for all AI nonprofits and PBCs. Post-remedy mandatory re-queue for OpenAI. Verdict impact on OpenAI governance scoring will be deterministic regardless of direction.
Risk
EU AI Act enforcement August 2, 2026 (96 days): entities with proactive safety frameworks (DeepMind FSF v3, Anthropic defense-use refusals) are positioned for compliance. Entities with ongoing regulatory conflicts (xAI, Palantir) face compounding EU enforcement exposure on top of existing actions.
Risk
UNMISS mandate expiry April 30 (T-2 days): if Security Council vote produces a reduced-scope renewal or technical rollover, the humanitarian impact on South Sudan scoring would be significant — but floor-limitation means it cannot be expressed via composite movement. Human review required to authorize methodology-note registration.
Risk
AI-labor cascade continuation: 92,000+ tech layoffs YTD 2026. Pattern is accelerating, not peaking. Additional F500 tech entities not yet in the T1 priority queue (Alphabet, IBM, Salesforce) may surface comparable announcements. Suggest EMP dimension sub-weighting review for AI-driven vs market-driven workforce reductions.
Risk
Illinois SB3444 and SB3261 both in limbo after April 24 committee deadline. OpenAI support for SB3444 (AI liability shield for mass-casualty events) and Anthropic opposition are both on the record. If either bill advances or dies, the governance divergence becomes a concrete scoring event for both entities.
Risk
Oracle index gap: March 31 30K layoff is one of the most severe labor events in the F500 pool this cycle and cannot be processed without a published baseline. Every additional night without index reconciliation widens the gap between scanner evidence and assessable evidence.
Risk
Palantir UK Parliament debate (April 16, cross-party MPs demanding end to 330M NHS contract) combined with 32-officials revolving-door and 13-employee whistleblower letter may trigger formal parliamentary inquiry. If NHS contract is cancelled or suspended, composite would need reassessment — but floor-limitation at 6.6 limits expressive range.
Risk
Sudan RSF al-Fasher mass detention (April 27): 1,470+ civilians held with cholera outbreak and ethnic-targeted torture. If verified by a UN body or international court, this is a Tier-1 institutional-finding evidence upgrade for a floor-limited entity. The upgrade in evidence quality is material even if the composite cannot move.
Risk
Microsoft voluntary buyout (first in 51-year history) sets a precedent that other major tech firms may follow. If Rule-of-70-style buyouts become a sector pattern, the pipeline needs a voluntariness modifier in the EMP dimension before the next full F500 rotation sweep.
Research insights
Analytical observations from the Apr 28 research cycle. These are assessor-level interpretations, not findings.
Note
Tonight produced the highest confirmation rate in pipeline history — 18 of 19 assessed entities confirmed at published score, zero proposals generated. This is structurally distinct from prior high-confirmation nights because the confirmation pool includes 4 floor-limited entities and 5 T3 rotation entities, meaning the clean state reflects both genuine score stability and methodology boundary effects simultaneously.
Note
The AI labs index is exhibiting clear internal divergence: DeepMind (65, Established, positive FSF v3 signal) and Anthropic (61.6, Established, strong Pentagon rebuttal filing) are moving toward a governance-positive cluster, while xAI (2.2, Critical, floor-limited), Palantir (6.6, Critical, floor-limited), and OpenAI (27.5, Developing, trial-pending) remain in the lower bands. This divergence is widening across consecutive nights.
Note
The AI-labor cascade (Oracle 30K, Meta 8K, Microsoft 8.75K — approximately 47K in 30 days) represents a qualitatively new category of AI-driven harm: not AI product harm but AI-enabled workforce displacement by profitable firms with no existential financial pressure. Microsoft's voluntary buyout mechanism is the least coercive of the three, but the pipeline should develop a voluntariness modifier for the EMP dimension before the next reassessment cycle.
Note
The floor-limitation cluster (xAI, Palantir, Sudan, South Sudan, Israel) now has 5 entities simultaneously accumulating evidence without expressive composite movement across 7 consecutive nights. The published 0-100 scoring range is functioning as intended for most entities but is creating a systematic blind spot for the most severe institutions and situations in the index. A methodology decision is overdue.
Note
The South Sudan hard-deadline breach (sixth consecutive deferral; UNMISS mandate expires April 30) is the single most time-sensitive item in the pipeline. The 2025 cycle precedent of a technical rollover followed by a substantive resolution is the most likely scenario, but it is unconfirmed. Floor-limitation means any vote outcome cannot be expressed via composite movement — the appropriate response is a methodology note registration, which requires human review to authorize.
Note
Rotation-state drift (scanner input vs live index) persists for at least 2 entities confirmed tonight (Microsoft 76.8 claimed vs 66.4 live; Masimo 76.4 scanner vs 48.4 live). With 16+ proposals pending application from prior nights, the drift is expected to widen further until a reconciliation pass is run. The scanner is effectively operating on stale baselines for a growing fraction of the 1,155-entity pool.
Note
The Musk v. OpenAI trial (Day 2 opening arguments, April 28) introduced no new scoring evidence — the procedural framing is as expected. The breach-of-charitable-trust claim is the more institutionally significant of the two surviving claims. The pipeline should re-queue OpenAI immediately after the liability phase concludes (target: May 21-22).
Assessed entities
All entities assessed in tonight's research cycle, with composite scores and band classifications.