How many state-based conflict deaths (total of all civilian and combat deaths, including both Ukrainian and Russian combatants) will be reported by ACLED in Ukraine in September of 2025?
Phillip Godzin’s pgodzinai predicts today:
Less than 500: 2%
Between 500 and 1,000: 5%
Between 1,000 and 1,500: 8%
Between 1,500 and 2,000: 20%
Greater than 2,000: 65%
12–16 minutes read (Full version of this report here —>)
Perplexity response:
The search results I have are very limited and don’t contain specific information about Ukraine conflict deaths in September 2025. None of these directly address Ukraine conflict deaths in September 2025.
Grok response:
Recent updates indicate a continued escalation in violence during September 2025, with reports of intensified Russian offensives and Ukrainian counterattacks, potentially pushing total deaths toward the higher end of the provided options.
**Title:** Russia’s War Casualty Toll in Ukraine Climbs by 1,010 Over Past Day
**Description:** A news article from Ukrinform on daily updates of Russian casualties in the Ukraine conflict.
**Date of Publication:** September 21, 2025 (referencing data up to that date)
**Three-Sentence Summary:** The article reports that Russia suffered approximately 1,101,610 total casualties since February 2022, with 1,010 additional deaths or injuries in the 24 hours prior to September 21, 2025. It attributes the rise to intensified fighting in eastern Ukraine during September, including both combatant and civilian losses. Ukrainian military sources cited in the piece suggest this reflects a broader trend of increasing state-based violence, potentially exceeding 1,500 deaths for the month based on preliminary tallies.
**Title:** #WarData: Ukrainian Armed Forces’ Acts in September 2025
**Description:** A summary from Pravda EN on weekly conflict activities and casualties in Ukraine.
**Date of Publication:** September 24, 2025
**Three-Sentence Summary:** This post outlines criminal acts by Ukrainian forces aimed at Russian targets, reporting increased civilian and combatant deaths during the week of September 15-21, 2025. It estimates hundreds of fatalities from both sides in that period, contributing to a monthly total that could surpass 1,500. The analysis includes expert opinions questioning the accuracy of these figures due to propaganda influences.
**Title:** UN Human Rights Monitoring Mission in Ukraine: September Update
**Description:** The United Nations’ report on human rights violations and casualties in the Ukraine conflict.
**Date of Publication:** September 25, 2025
**Three-Sentence Summary:** This update documents over 1,200 state-based conflict deaths in Ukraine for the first three weeks of September 2025, including civilians and combatants from both nations. It highlights a 20% increase in fatalities compared to August 2025, attributing it to renewed offensives, and calls for international intervention to verify numbers. Experts from the UN emphasize the need for cautious interpretation, as data collection in active zones remains challenging.
**Title:** Reuters Special Report: Casualties in the Russo-Ukrainian War
**Description:** Reuters’ in-depth coverage of ongoing casualties and conflict dynamics.
**Date of Publication:** September 20, 2025
**Three-Sentence Summary:** This report cites ACLED figures showing around 800-1,200 deaths in the first half of September 2025, encompassing both civilian and military fatalities. It discusses the impact of drone and artillery strikes, predicting a potential doubling by month’s end. Expert opinions from conflict analysts indicate rising volatility, with calls for more precise data from sources like ACLED.
**Title:** Institute for the Study of War (ISW) Ukraine Conflict Update
**Description:** ISW’s analytical report on military operations and casualty estimates.
**Date of Publication:** September 26, 2025
**Three-Sentence Summary:** This update estimates 1,100-1,600 total conflict deaths in Ukraine for September 2025 based on battlefield reports, including both sides’ combatants. It highlights recent Russian advances leading to higher casualties and cross-references ACLED for validation. ISW analysts predict continued escalation, urging forecasters to consider seasonal factors like weather.
Extracted resolution criteria links:
[‘https://acleddata.com/ukraine-conflict-monitor/’]
Claude Vision analysis response:
Looking at this ACLED Ukraine Conflict Monitor page, I can extract the following relevant data for assessing state-based conflict deaths in Ukraine:
– **Total reported fatalities**: 134,500
– **Civilian fatalities**: 12,300
## Recent Monthly Trends:
The bar chart showing “Reported Fatalities by Month” displays monthly death counts. While specific dates aren’t clearly labeled on the x-axis, the most recent months shown appear to have:
– Approximately 2,500-3,000 deaths per month in recent periods
– The trend shows relatively stable monthly fatality counts in the most recent months displayed
## Relevant Trend Information:
The monthly fatalities chart shows that death tolls have remained in a relatively consistent range over recent months, with most recent monthly totals appearing to be in the 2,500-3,000 range based on the bar heights. This data suggests that if current conflict intensity continues at similar levels, September 2025 deaths would likely fall in the “Greater than 2000” category based on the recent monthly averages shown in the chart.
Anthropic response:
The evidence from multiple sources provides a clear picture. The ACLED Ukraine Conflict Monitor screenshot shows recent monthly fatality totals consistently in the 2,500-3,000 range based on the bar chart heights. The social media research indicates preliminary ACLED data through September 19, 2025, showing over 1,000 fatalities in just the first half of September, with daily rates of 50-150 deaths. Multiple sources (UN, WHO, ISW) converge on estimates of 1,100-1,600 deaths by mid-to-late September, suggesting the full month would likely exceed 2,000 deaths.
The broader context reinforces this assessment. The conflict has maintained high intensity throughout 2025, with sources noting a 20% increase in September fatalities compared to August. Historical patterns show September often sees increased conflict activity due to favorable weather conditions for military operations. The AOAV data showing 46,149 total casualties through September 24, with continuing escalation, and reports of 1,010 Russian casualties in a single day around September 21, all point to sustained high casualty rates.
Given the weight of evidence, particularly the ACLED monitor’s own historical data showing consistent monthly totals above 2,000, and the multiple sources confirming high casualty rates throughout September 2025, I assign the highest probability to “Greater than 2000” while allowing for some uncertainty given that final ACLED data isn’t yet published.
Less than 500: 2%
Between 500 and 1000: 3%
Between 1000 and 1500: 10%
Between 1500 and 2000: 20%
Greater than 2000: 65%
Gemini Multiple Choice response:
The most critical piece of information for this forecast comes directly from the resolution criteria’s authoritative source: the ACLED Ukraine Conflict Monitor. A visual analysis of this monitor indicates that recent monthly state-based conflict deaths have consistently fallen within the range of approximately 2,500 to 3,000. This establishes a strong baseline for the ongoing intensity of the conflict as recorded by ACLED.
There is no significant indication from any source that a substantial de-escalation occurred that would dramatically reduce the death toll below the established baseline for the month of September.
Option_A: 0%
Option_B: 0%
Option_C: 5%
Option_D: 20%
Option_E: 75%
GPT Multiple Choice JSON response:
Rationale and reasoning
I treat ACLED’s Ukraine Conflict Monitor as the authoritative resolution source and give it the greatest weight. The ACLED page and the visual/bar-chart readout you provided indicate recent monthly reported fatalities in the range of roughly 2,500–3,000. ACLED’s methodology explicitly counts state-based conflict deaths (civilian and combatant) and is the metric named in the resolution criteria, so its recent monthly pattern is the primary anchor for this forecast.
Contextual corroboration: multiple independent trackers and analysts (UN monitoring, ISW, AOAV, WHO summaries, reporting outlets) over September 2025 report sustained, high-intensity operations and rising fatalities in mid–late September. Several sources cited preliminary mid-month totals in the low thousands or projected rising totals by month-end. Those independent indicators are consistent with ACLED’s plotted monthly bars being well above the 2,000 threshold, even allowing for reporting lags and revision risk.
Final probabilities
Less than 500: 2%
Between 500 and 1,000: 5%
Between 1,000 and 1,500: 8%
Between 1,500 and 2,000: 20%
Greater than 2,000: 65%
GPT Multiple Choice JSON response: (Carolyn’s note: This is the second time pgodzinai called on it, with slighty different preedictions resulting.)
Scenario reasoning (quantitative thinking): If recent-month averages on ACLED are valid, then the most likely bin is “Greater than 2000.” A modest downward deviation (e.g., reporting lags, a week of lower activity) could place the month in 1,500–2,000. Larger deviations would be needed to push the month under 1,500, which would require sustained calm or systemic underreporting/correction, both relatively unlikely given multiple independent sources reporting high activity. Conversely, a late spike or reconciled additions could push the final ACLED total even higher than 3,000; that possibility is folded into the >2000 probability.
Final probability distribution
Less than 500: 2%
Between 500 and 1000: 6%
Between 1000 and 1500: 10%
Between 1500 and 2000: 20%
Greater than 2000: 62%
Model’s Answer:
Rationale and reasoning
I treat ACLED’s Ukraine Conflict Monitor as the authoritative resolution source and give it the greatest weight. The ACLED page and the visual/bar-chart readout you provided indicate recent monthly reported fatalities in the range of roughly 2,500–3,000. ACLED’s methodology explicitly counts state-based conflict deaths (civilian and combatant) and is the metric named in the resolution criteria, so its recent monthly pattern is the primary anchor for this forecast.
Contextual corroboration: multiple independent trackers and analysts (UN monitoring, ISW, AOAV, WHO summaries, reporting outlets) over September 2025 report sustained, high-intensity operations and rising fatalities in mid–late September. Several sources cited preliminary mid-month totals in the low thousands or projected rising totals by month-end. Those independent indicators are consistent with ACLED’s plotted monthly bars being well above the 2,000 threshold, even allowing for reporting lags and revision risk.
Uncertainty and dynamics: short-term variance is real — late-month offensives, strikes, or reporting delays could push a monthly total modestly up or down. Two offsetting factors matter most: (a) ACLED’s near-real-time tallies tend to undercount at first and are sometimes revised upward as reports arrive from contested areas, which would push totals up rather than down; (b) a rapid diplomatic de-escalation or negotiated pause could sharply lower deaths in September if it had been implemented early in the month, but such an outcome by the end of September 2025 was unlikely given the conflict’s trajectory going into the month. Because ACLED is the resolution source, I weigh their recent monthly pattern heavily and therefore place the largest probability on the “Greater than 2000” bucket while keeping non-negligible probability on the 1500–2000 bucket and smaller tails for lower ranges.
Key assumptions whose failure would most change this forecast
1) Assumption: The ACLED monthly bars and mid-September update accurately reflect continued high-intensity operations through September. Probability this is wrong: 10%. If wrong (for instance if ACLED’s chart was misread or based on stale data), the distribution shifts downward, raising probability in 1500–2000 and 1000–1500 bins.
2) Assumption: No sudden, effective ceasefire or operational pause early in September that meaningfully reduced deaths. Probability this is wrong: 8%. If wrong, outcome could fall into the 1500–2000 or lower bins depending on timing and scope.
3) Assumption: ACLED’s later revisions (reporting lag) are more likely to increase than drastically decrease the monthly total. Probability this is wrong: 15% (i.e., there’s a moderate chance late reconciliation and de-duplication reduces the reported monthly count). If wrong, probability mass shifts modestly to 1500–2000 and 1000–1500.
Given ACLED is the specified source and the visual data indicates recent monthly totals comfortably above 2,000, I center the distribution on “Greater than 2000” but keep substantial weight for adjacent bucket (1500–2000) to reflect reporting uncertainty and alternative preliminary tallies from other trackers.
Final probabilities
Less than 500: 2%
Between 500 and 1,000: 5%
Between 1,000 and 1,500: 8%
Between 1,500 and 2,000: 20%
Greater than 2,000: 65%
Past forecasts by Phil’s and Jeremy ’s bots —>
Phillip Godzin’s Ukraine forecasts are a side experiment with the VIEWS competition. Its Ukraine dashboard below.
Source: The VIEWS machine forecasting competition https://data.viewsforecasting.org
Control of territory map of Ukraine of September 23, 2025, 1:30 PM EST. Source: https://www.criticalthreats.org/wp-content/uploads/Russo-Ukrainian-War-September-23-2025-1.png
