What moved
Attacks claimed by the Houthis kill three at Saudi airports, and Saudi Arabia strikes back
Attacks that Yemen's Houthis say they carried out killed three people at two of Saudi Arabia's international airports this week. Saudi Arabia's civil aviation authority said on Wednesday 7 October that Abha International Airport and King Khalid International Airport, in the capital, Riyadh, had been attacked on Tuesday and Wednesday, killing three people and wounding 36; it blamed no one (The Straits Times, carrying AFP). By the authority's breakdown as the BBC and Reuters give it, two people were killed and 28 wounded at Abha, near the Yemeni border, and one killed and eight wounded in Riyadh; it did not say which airport was hit on which day (BBC; The Korea Herald, carrying Reuters). The Houthis, the movement that holds most of Yemen's north-west, said they had attacked both airports with ballistic missiles and drones (BBC). Euronews describes the attack that killed one person at the Riyadh airport as a drone attack (Euronews). These were not the first attacks on the kingdom: since September, by AFP's account, the Houthis have regularly targeted oil and military installations in Saudi Arabia (The Straits Times, carrying AFP). Yemen has 40.6 million people and Saudi Arabia 35.3 million (World Bank, 2024; World Bank, 2024).
Our sources do not agree on who was killed at Abha, though the authority's own statement does. The statement, as carried by Qatar's news agency QNA and the Saudi outlet Sabq, says the two killed at Abha were residents from Morocco and Algeria (QNA; Sabq). The BBC identifies them as two women, a Moroccan and an Algerian, and Euronews as Moroccan and Algerian residents (BBC; Euronews). AFP's report, as carried by The Straits Times, Dawn and France 24, has an Egyptian and an Algerian, and so does Middle East Eye (The Straits Times, carrying AFP; Dawn; France 24; Middle East Eye). The sources that name the person killed in Riyadh all say he was Sudanese. Five Saudi citizens, two Egyptians and a Pakistani were among those hurt in Riyadh, the authority said (Dawn). QNA's version of the statement gives 30 hurt at Abha, 28 of them Saudi citizens and two residents from Bangladesh and the Philippines; with the eight in Riyadh that makes 38, against the 36 in the same statement (QNA). The BBC and Reuters give 28 at Abha.
The claims went on through the window. The Houthis said they had struck the Riyadh airport on the night of Monday 5 October (The Straits Times, carrying AFP). The Houthi military spokesman, Yahya Saree, said on Wednesday that a ballistic missile had hit King Khalid airport and "led to a disruption of air traffic"; the Saudi-led coalition said that missile was intercepted (Dawn). Our sources do not say whether it was one of the attacks that killed people. On Thursday Saree claimed another missile strike on the airport at midday. The coalition said it had intercepted two missiles fired at Riyadh that day, and the Saudi civil defence agency reported "material damage" to buildings from falling fragments (Euronews). AFP journalists heard several explosions across the city on Thursday (France 24). Diplomats were told to shelter in their embassies, and the airport asked departing passengers to check with their airlines before setting out (Euronews). On Thursday Saree also told staff at Saudi oil facilities to leave "areas that are targets for our forces"; in an earlier post he had warned airlines "for the last time" against flying in Saudi airspace, except over Mecca and Medina (Euronews).
On the same Wednesday, Yemen's government said at least two Houthi ballistic missiles and several drones were fired at the airport in Aden, the city where it sits; one came down close to the runway shortly before a Cairo flight was due to land, and the flight was sent on to Jeddah (Al Jazeera). The Houthis said they had aimed at Saudi military supplies at a camp near the airport (Al Jazeera).
Saudi Arabia answered. The coalition it leads said the attacks on the airports "will not go unpunished" and that it had struck 82 "legitimate military targets", among them missile sites, command centres and gathering points (The Straits Times, carrying AFP). Reuters reports the figure as more than 80 sites, in the governorates of Saada, Hodeidah, Jawf and Marib (The Korea Herald, carrying Reuters). On Wednesday the coalition said it was tracking the platform in Sanaa from which a missile was launched at Riyadh and would destroy it; it later said it had done so (The Korea Herald, carrying Reuters; Euronews).
How the war reached this point, as our sources tell it. Yemen's civil war broke out again in July after a four-year ceasefire, and in September, by AFP's account, the Houthis seized the entire western coastline along the Red Sea (The Straits Times, carrying AFP); the BBC says "strategically important stretches" of it (BBC). They took the port of Mocha in September (The Straits Times, carrying AFP). That coast runs up to the Bab el-Mandeb strait, at the southern end of the Red Sea and its narrowest point (BBC), which Reuters describes as more important for energy exports since the war between the United States and Iran restricted traffic through the Strait of Hormuz (The Korea Herald, carrying Reuters). The government announced on Sunday 4 October that its aim was to retake all the territory it has lost to the Houthis, and began a counter-offensive; on Monday the Saudi-led coalition said it was providing air support (Middle East Eye; The Straits Times, carrying AFP).
What that counter-offensive has achieved is disputed. Government forces claimed to have pushed the Houthis out of the area around the strait and Mocha; AFP dates the claim Monday and Middle East Eye Tuesday, and AFP adds that military sources with the government would not officially confirm taking Mocha, saying only that they had cut its one supply route (The Straits Times, carrying AFP; Middle East Eye). The Houthis denied it and posted videos of their fighters inside Mocha; Middle East Eye says other reports and open-source material appeared to support them (Middle East Eye). Reuters cites Yemeni military officials on the ground and analysts as saying the government's claim to have reached Mocha was premature (The Korea Herald, carrying Reuters). DW, by contrast, reports that the operation drove the Houthis out of the areas around the strait and Mocha (DW). AFP says military sources on both sides reported no major change on the ground, and that neither side's claims can be independently verified because the situation moves fast and journalists have little access (The Straits Times, carrying AFP). On Wednesday, AFP was told by military sources from both camps, at least 62 fighters were killed in new clashes in the west, 37 of them Houthis (Dawn).
Other governments are being drawn in. Pakistan and Turkey agreed to activate measures under their defence pact with Saudi Arabia and to send forces to the kingdom; Al Jazeera dates the agreement Monday and Middle East Eye Tuesday (Al Jazeera; Middle East Eye). Two officials told Reuters that Turkey is sending mainly defensive and technical support. A US official and a Syrian military official told Reuters that Syria is considering military help, and the US official said Saudi Arabia had asked Syria to send forces; a Syrian diplomat said Syria stood firmly with Saudi Arabia but that its government's primary focus was on domestic priorities such as recovery from its civil war (The Korea Herald, carrying Reuters). An analysis in the South China Morning Post says the US military has reportedly sent more than 100 advisers to Riyadh, that Britain has agreed to refuel Saudi aircraft in flight, and that French air-defence units are defending Saudi oil-export infrastructure (South China Morning Post). Saudi Arabia's coalition accuses Iran of arming, training and funding the Houthis; the Houthis deny being part of Iran's network, and Iran rejects the accusation (The Korea Herald, carrying Reuters).
The cost to people in Yemen is the part of this with the largest numbers. More than 200,000 people have been displaced since the war escalated again in July, according to UN data cited by Euronews (Euronews). AFP attributes that figure to the United Nations; Reuters and DW attribute it to the International Organization for Migration, a UN agency, which DW says counted about 76,000 in mid-September (France 24; The Korea Herald, carrying Reuters; DW). An AFP report the day before the 200,000 figure was given put it at more than 184,000 (The Straits Times, carrying AFP). The organisation's director general, Amy Pope, said people were "running for their lives" (The Korea Herald, carrying Reuters). In parts of Yemen, by the UN's figure, three families in every four are going hungry, and each side says hundreds of people have been killed, without giving details (The Korea Herald, carrying Reuters).
The EU's trade chief goes to Beijing to head off a trade war
Maroš Šefčovič, the European Commission member responsible for trade, went to Beijing for two days of talks with China's commerce minister, Wang Wentao (Euronews). Euronews reported on Thursday that he had landed the day before; our sources do not report what came of the talks within the window, so what follows is what each side brought to them (Euronews). The dates differ between sources: Euronews on 7 and 8 October has him travelling on Wednesday, while Euronews on 6 October and The Japan Times said Thursday (Euronews; The Japan Times). The European Union has 450.2 million people and China 1,409.0 million (World Bank, 2024; World Bank, 2024).
The European side says the imbalance is the problem. Euronews puts the EU's trade deficit with China at a record €1 billion a day, and reports that every EU country now runs a deficit with China (Euronews; Euronews). The Japan Times gives the deficit for last year as around €360 billion, and says China puts it lower, at around $292 billion, and expects it to rise this year (The Japan Times). The two figures are in different currencies, as the source gives them, and are not converted here. The EU's industry commissioner, Stéphane Séjourné, told Euronews last month that the EU was "losing thousands of jobs every week" and had lost 250,000 industrial jobs last year (Euronews).
Šefčovič has named three main objectives; the report we read gives two of them, curbing surges of Chinese exports into the EU, especially in critical sectors, and selling more to China. Brussels also wants "a system of export licensing for rare earths and other products" (The Japan Times). Rare earths are the EU's weak point: China controls their production and processing, and restricted exports of them in 2025 during its dispute with the United States (Euronews). Brussels wants quotas, product by product, on Chinese goods (Euronews). China has rejected a proposal for "voluntary export restrictions" on electric vehicles: on 20 September a Chinese spokesperson called such measures contrary to the "principle of fair competition" and said any solution must "comply with WTO rules" and suit both sides' industries (Euronews). The Commission had given Beijing until October to deliver "tangible" results (Euronews). People familiar with the talks told the South China Morning Post that Brussels hoped China would agree, by the end of Friday, to moderate its exports in a single sector as a "proof of concept" (South China Morning Post).
The talks come with threats on both sides. On Monday, Germany's chancellor, Friedrich Merz, and France's president, Emmanuel Macron, wrote to the Commission's president, Ursula von der Leyen, asking for a new instrument that would allow "powerful measures up to an immediate cut-off from the internal market if needed" (South China Morning Post). In the Post's reading, the paper the two countries co-signed marks a dramatic step-change for a German government long reluctant to confront Beijing on trade; Euronews notes that Germany opposed EU tariffs on Chinese electric vehicles in 2024 (South China Morning Post; Euronews). China's commerce ministry said on Tuesday that it hoped France and Germany would not "politicise or oversecuritise economic and trade issues" (South China Morning Post). On Saturday 3 October China opened an anti-dumping investigation into a chemical imported from the EU, after an EU investigation in September into imports of polyvinyl chloride from countries including China; Euronews reports the move as seen as retaliation for EU investigations into Chinese chemical producers (Euronews; Euronews).
Few in our sources expect much. "There may be a few crumbs, but I would not expect any kind of major breakthrough," said Penny Naas of the German Marshall Fund (The Japan Times). Šefčovič told Euronews in an interview in early September that a failure to deliver tangible results would bring "a strong political movement" among EU leaders for "harsher measures"; those leaders meet in Brussels next week (Euronews).
The International Criminal Court unseals warrants for four Taliban leaders over the persecution of women and girls
On Monday 5 October, judges at the International Criminal Court in The Hague made public four arrest warrants against senior figures of Afghanistan's Taliban government (UN News; The Hindu). Two were already known to exist: those for the Taliban's supreme leader, Haibatullah Akhundzada, and the head of its Supreme Court, Abdul Hakim Haqqani, issued on 8 July 2025 under seal (UN News). Two had not been made known: for the minister of education, Habibullah Agha, and the minister of higher education, Neda Mohammad Nadim, issued on 9 September 2025 (Dawn, carrying AFP; Afghanistan International). Some reports describe Monday's step as further detail on warrants already issued; all agree that the four were unsealed (The Express Tribune; Anadolu). Afghanistan has 42.6 million people (World Bank, 2024).
The judges found reasonable grounds to believe that each of the four ordered, induced or solicited persecution, which the court's founding treaty counts as a crime against humanity, on two grounds: gender, against girls, women and others who did not conform to Taliban policy on gender, gender identity or expression; and politics, against people seen as "allies of girls and women" (UN News). In their finding, Taliban policy led to serious violations of fundamental rights and freedoms, connected with murder, imprisonment, torture, rape and enforced disappearance, from the takeover on 15 August 2021 until at least 20 January 2025 (UN News). That end date is for the warrants against Akhundzada and Haqqani; the ministers' warrants cover suspected crimes through at least 6 March 2025 (Courthouse News). The court's Pre-Trial Chamber II said it unsealed the warrants, with parts blacked out, after concluding that this would not unduly affect investigations, and that publication could help prevent further crimes by the suspects (South China Morning Post, carrying AFP).
The two ministers are accused over the policies on schooling. Our sources describe the school ban in three ways. AFP says girls are barred from school beyond primary level; the version of its report carried by the South China Morning Post says girls over 12; The Hindu says women are barred from education beyond sixth grade (Dawn, carrying AFP; South China Morning Post, carrying AFP; The Hindu). AFP and UN Women say women are barred from university. UN Women says Afghanistan is the only country in the world that bars girls from secondary school and women from university, and UN figures say half of Afghan women now leave home only once or twice a month (The Express Tribune). Health workers told AFP that hospitals are short of staff because women can no longer qualify as doctors (Dawn, carrying AFP).
The Taliban government rejected the court. "We are bound by the rulings of Islamic sharia and its courts, not by intelligence entities or so-called courts," its chief spokesman, Zabihullah Mujahid, said (South China Morning Post, carrying AFP). The court has no police of its own, and with the Taliban in control of the country The Hindu calls the prospect of any arrest remote (The Hindu). On the same day, the UN Human Rights Council urged the Taliban to repeal what it called their "worsening" and "systematic oppression" of Afghan women and girls, and its 47 member countries unanimously criticised that discrimination (South China Morning Post, carrying AFP).
Ebola reaches Kenya: a man who returned from Congo dies in Nairobi
Kenya confirmed its first case of Ebola on Tuesday 6 October, and the patient was already dead. The health minister, Aden Duale, said the man, a Kenyan citizen who had lived in the Democratic Republic of Congo for seven years, fell ill about a month earlier and was treated in several hospitals there (BBC). He travelled by road to Kampala, in Uganda, and flew to Nairobi on Saturday 3 October, passing through the normal health screening at the airport (The EastAfrican). A relative and a friend drove him to Nairobi Hospital, where he was isolated; he tested positive for the Bundibugyo species of Ebola at two laboratories and died at about 23:30 on Monday (The EastAfrican; The Guardian). He was buried on Tuesday under safe-burial rules (PBS, carrying the Associated Press). Kenya has 56.4 million people (World Bank, 2024).
How he passed screening is a question the ministry raised itself. Patrick Amoth, whom our sources call the ministry's director general or acting director general, said he may have got through because medication masked his symptoms (Daily Maverick; AllAfrica, carrying Capital FM). "The failure is the system that did not work," Amoth told reporters, asking how someone so obviously ill could travel (South China Morning Post). The ministry has identified 28 contacts, among them family members and health workers, and is tracing 23 passengers and four crew members from his flight (The Guardian). At the time of the announcement only one test, his, was positive (Dawn). Contacts are to be isolated for up to 21 days and released after testing negative (PBS, carrying the Associated Press; AllAfrica, carrying Capital FM). Kenya's doctors' union called for health workers who may have been exposed to be identified and monitored at once: "Our healthcare workers must not be the weakest link in our response" (AllAfrica, carrying Capital FM).
How many countries the outbreak has now reached is counted two ways. The World Health Organization and the Associated Press count Kenya as the fourth country with a case in this outbreak, after Congo, Uganda, which recorded 20 cases and two deaths before being declared free of the disease in August, and France, which had one case linked to travel in June (Premium Times; PBS, carrying the Associated Press; Dawn). Daily Maverick calls Kenya the third (Daily Maverick). One report has Uganda declared free in July, not August (AllAfrica, carrying Capital FM).
In Congo itself the outbreak is still growing. The WHO had recorded 4,082 deaths among 8,463 cases as of Friday 2 October (South China Morning Post; The Guardian); the BBC, citing the WHO, gives 4,114 deaths among 8,544 cases (BBC). Congo has 109.3 million people (World Bank, 2024). Médecins Sans Frontières said on Monday that North Kivu province, which borders Uganda, now accounts for 40% of new cases, against 24% at the end of August; the BBC gives it as nearly 40%. "It is like fighting a megafire," said Stephanie Hoffmann, who coordinates its treatment centre in Butembo (The Guardian; BBC). There is no approved vaccine or treatment for the Bundibugyo species (BBC).
Nigeria's disease-control agency said the case showed the risk of the virus travelling, including to Nigeria, which had no confirmed case as of 6 October (Premium Times). The WHO said it had sent about 1,000 tests and 1,000 sets of protective equipment to Kenya's counties most at risk (PBS, carrying the Associated Press).
Imran Khan's party marches on Islamabad, and the prime minister offers talks if it stops
Supporters of Pakistan's jailed former prime minister, Imran Khan, set out on Sunday 4 October from Lakki Marwat, in the north-western province of Khyber Pakhtunkhwa, on a march to demand his release that our sources expect to head for the capital, Islamabad (NPR; Dawn). The BBC notes that the organisers had not announced Islamabad as their destination, though the authorities blocked the routes into it (BBC). The BBC put the crowd at the start at some 5,000, and NPR at more than 5,000 (BBC; NPR). It is led by the province's chief minister, Sohail Afridi; the province is governed by Khan's party, Pakistan Tehreek-e-Insaf (The Guardian; BBC). Pakistan has 251.3 million people (World Bank, 2024). The previous issue passed this story over because the march had only just begun; it was still under way on Tuesday.
The march went ahead because talks failed. The party and the government talked on Friday and Saturday, and the talks collapsed early on Sunday (Dawn; The Hindu). Most of our sources count two rounds; Geo News counts three (Geo News). Each side blames the other. The party says the sticking points were access to Khan, whom its leaders want to consult, and the government's demand that the provincial government back its counter-terrorism measures in the province (Dawn; The Express Tribune). The interior minister, Mohsin Naqvi, accused Afridi of "sabotaging" the negotiations (Dawn). A junior interior minister, Tallal Chaudhry, said a draft agreement had been ready by 9pm on the second day until the party's delegation stepped out to consult Afridi and came back asking to leave (Dawn). The party denies that matters had been settled, and says Afridi had no role in the deadlock (Geo News; The Express Tribune).
On Tuesday the prime minister, Shehbaz Sharif, made an offer with a condition. He asked the party to call off the march first: "When you end them, we will surely sit down with you." Sharif warned that the march would hurt an economy that he said had reached "macroeconomic stability" (Dawn). The party's chairman, Gohar Khan, called the offer "merely symbolic" in Geo News's rendering, and "a mere offer" in the Express Tribune's, and said its leaders had not been allowed to meet Khan since October 2025 (Geo News; The Express Tribune). On Wednesday Mahmood Khan Achakzai, who leads the opposition in the National Assembly, said: "Until I meet Imran Khan, I will not be part of any negotiations" (Dawn).
The state's response has been to block the road and detain relatives. Shipping containers have been stacked across the routes into Islamabad (BBC). Our sources disagree on when Khan's sisters were detained. The Guardian says all three were among those detained in the week before the march, on grounds of national security; the BBC reports that one of them, Aleema Khanum, was arrested on 20 September, two weeks before it set out, for, in the arrest order's words, mobilising people to "create law and order situations" (The Guardian; BBC). The party's chairman said on Tuesday that a cousin, Qasim Zaman, had also been detained (Geo News; The Express Tribune). A government source told Dawn that the government was seriously considering imposing governor's rule in the province, and the party warned that this would have serious consequences (Dawn; The Express Tribune). By Tuesday the march had reached Kohat, still inside Khyber Pakhtunkhwa (Geo News). Our sources do not say where it was when the window closed on Thursday.
Khan, 74, was first imprisoned in August 2023, and was sentenced last year to 17 years for corruption; he says all the cases against him are politically motivated (BBC). His family says he has lost most of the sight in his right eye in prison (BBC). Officials say his health is good and that his family visits him regularly, but that meetings with him cannot be used for political purposes (NPR); his family and party colleagues say he has been denied visits by his doctor and relatives (The Guardian).
Still happening
Brazil's run-off: the first polls put Flávio Bolsonaro narrowly ahead of Lula
The previous issue reported the first round of Brazil's presidential election. Flávio Bolsonaro, a right-wing senator of the Liberal Party, took 47.03% of valid votes against 45.16% for the president, Luiz Inácio Lula da Silva, and the two meet in a run-off on 25 October (Anadolu; Al Jazeera). Brazil has 212.0 million people (World Bank, 2024). What changed in this window is the first polls of the second round and a run of endorsements.
The first two polls point the same way by different margins. Datafolha, which interviewed voters in 153 cities on Tuesday and Wednesday for the newspaper Folha de S.Paulo and TV Globo, has Bolsonaro on 49% and Lula on 45%; counting valid votes only, as the electoral court does, 52% to 48%. Its margin of error is two points either way, so the two are tied at the very edge of it (G1). Folha adds that such a tie is statistically unlikely and points to an advantage for Bolsonaro (Folha de S.Paulo). PoderData/Aya, which interviewed 3,000 people by phone from Monday to Wednesday, has 53% to 47%, with a margin of error of 1.8 points; its last poll before the first round had the two level at 50% (Anadolu). Datafolha finds 46% of voters saying they would not vote for Lula and 43% saying so of Bolsonaro (Folha de S.Paulo).
Endorsements have gone mostly one way. The Republicanos party backed Bolsonaro on Wednesday, without obliging its members to follow, and he already had the support of a two-party federation that elected 87 deputies (Folha de S.Paulo). Two defeated candidates, Ronaldo Caiado of the PSD and Romeu Zema of Novo, have backed him, though Caiado's party stayed neutral (Folha de S.Paulo). Al Jazeera reported on Monday that Caiado had said he would endorse no one; by Thursday Reuters reported his endorsement as made earlier in the week (Al Jazeera; The Straits Times, carrying Reuters). On Thursday Bolsonaro announced the endorsement of Augusto Cury, who came third with 2.89% of valid votes, after local media had reported it (The Straits Times, carrying Reuters). Datafolha finds 46% of those who voted for four of the eliminated candidates moving to Bolsonaro and 32% to Lula (Folha de S.Paulo).
Lula's campaign is betting on the people who did not vote. Abstention in the first round was a record 21.08%, or 33.4 million voters, while Lula finished 2.2 million votes behind (Folha de S.Paulo). People close to the campaign told Reuters that Lula blames its messaging for failing to reach enough young people and women (The Straits Times, carrying Reuters). Folha notes that since Brazil's return to democracy no candidate who came second in the first round has gone on to win (Folha de S.Paulo). Lula's own team read the Datafolha numbers as a tie and the race as open; an aide to Bolsonaro said, in Portuguese, that they had gained more points than Lula (G1).
The slow story
The World Bank cuts its count of the world's poorest people, and most of the cut is one survey in one country
Our watchlist reads the World Bank's measure of extreme poverty before every issue. It flagged the measure this time because the figure for 2024 had changed since we last looked: our previous reading was 10.4% of the world's people, and the series now gives 10.1% (World Bank, World Development Indicators). The Bank's March release matches the earlier figure (World Bank, March 2026 release). The line is $3.00 a day per person, at 2021 prices adjusted for what money buys in each country. Our watchlist's own label still names the older line of $2.15 a day; the series it reads now uses the newer one.
The change comes from the Bank's September 2026 update to its Poverty and Inequality Platform. The Bank now estimates that 824.3 million people lived in extreme poverty in 2024, about one person in ten, down from 846.8 million in its March estimate for the same year: 22.5 million fewer people, a revision the Bank calls minor (World Bank, blog). The revision, in the Bank's words, comes "primarily" from a lower rate in South Asia, after the latest survey there replaced "the previously used extrapolation" (World Bank, blog). Its technical note names the survey: India's Household Consumption Expenditure Survey for 2023 to 2024 (World Bank, technical note).
| Region | 2024, March 2026 estimate | 2024, September 2026 estimate |
|---|---|---|
| World | 846.8 million (10.4%) | 824.3 million (10.1%) |
| South Asia | 63.4 million (3.8%) | 42.9 million (2.6%) |
| Sub-Saharan Africa | 582.1 million (45.1%) | 580.6 million (45.0%) |
| Middle East, North Africa, Afghanistan and Pakistan | 116.8 million (14.4%) | 116.4 million (14.3%) |
Source: the World Bank's March 2026 and September 2026 releases on its Poverty and Inequality Platform, read during this run; the same figures are in Table 2 of its technical note. People below $3.00 a day at 2021 prices, with the share of each region's population. The other regions barely changed; their figures are in the note.
So 20.5 million of the 22.5 million come from South Asia, which in the Bank's grouping does not include Pakistan or Afghanistan. For India, the note gives an extreme-poverty rate of 5.3% for the year it labels 2022 and, from the new survey of 2023 to 2024, 2.6% for the year it labels 2023 (World Bank, technical note). India has 1,450.9 million people (World Bank, 2024).
Confidence. The Bank itself says not to lean on that one-year fall. Its note calls the drop between the two survey rounds "unusually large", and says it "may partly reflect improvements in survey implementation and reporting, in addition to real welfare gains" (World Bank, technical note). It checked the obvious explanations and found that the change cannot be put down to how the sample was drawn or weighted, nor fully to cuts in food subsidies. Comparisons with labour-market figures and other data "support continued poverty reduction but suggest a more moderate pace than implied by the survey". The Bank recommends reading annual movements in India "with caution" and putting more weight on the long-term trend from 2011-12 to 2023-24; the note attributes that advice to a separate Bank paper on India (World Bank, technical note). The world's number moved, then, mainly because a projection for India was replaced by a survey. The Bank says the fall that survey shows likely reflects both real gains in welfare and better data collection in the survey's second year, and does not say in what proportion.
The revision changes little about where the world's poorest people live, and that is the slower story under it. The same release lets us set Sub-Saharan Africa against the world over 36 years:
| Year | World, below $3.00 a day | Sub-Saharan Africa, below $3.00 a day | Sub-Saharan Africa's share of the world total |
|---|---|---|---|
| 1990 | 2,300.6 million | 320.6 million | 13.9% |
| 2000 | 2,230.9 million | 425.3 million | 19.1% |
| 2010 | 1,469.1 million | 442.0 million | 30.1% |
| 2015 | 998.8 million | 465.1 million | 46.6% |
| 2020 | 896.3 million | 537.6 million | 60.0% |
| 2024 | 824.3 million | 580.6 million | 70.4% |
| 2026, the Bank's nowcast | 810.9 million | 586.2 million | 72.3% |
Source: our reading of the World Bank's September 2026 release, read during this run. The shares are our arithmetic on the Bank's figures. The figures for 2025 and 2026 are the Bank's nowcasts, built on national-accounts projections from its Global Economic Prospects of June 2026 (World Bank, blog).
In 1990 about one in seven of the world's people in extreme poverty lived in Sub-Saharan Africa. In 2024 it was seven in ten. The share of the region's own people below the line has fallen, from 61.5% in 1990 to 45.0% in 2024, but its population has grown faster, so the number has risen while the world's has fallen from 2,300.6 million to 824.3 million (World Bank, September 2026 release). The Bank's blog says extreme poverty is below 5% in every region except two: Sub-Saharan Africa, where it "remains high", and the region it calls the Middle East, North Africa, Afghanistan and Pakistan, where it is rising (World Bank, blog). Its note adds that survey coverage in Sub-Saharan Africa for 2024 is 65.57% of the population, against 98.66% for South Asia (World Bank, technical note). On our reading, the region where most of the poorest people live is also the one the Bank sees least well, though the Bank judges survey coverage sufficient in every region for 2022 to 2024 (World Bank, blog).
Pattern watch
- Hormuz and Bab el-Mandeb. From issue one. Read through the lead story: the fighting in Yemen has focused in recent days around the Bab el-Mandeb strait, which Reuters describes as more important for energy exports since traffic through the Strait of Hormuz was restricted; see above.
- Displacement out of Yemen. From issue one. Written up for the first time since issue one: more than 200,000 people displaced inside Yemen since July, by the UN or the International Organization for Migration depending on the source; see the lead story. Our sources describe displacement inside the country, not out of it.
- Ebola in the Democratic Republic of Congo. From issue three. The first case in Kenya, and North Kivu's share of new cases rising; see above.
- Brazil's presidential run-off. From the previous issue. The first two polls put Bolsonaro ahead; see above.
- Ethiopia and Tigray. From issue two. Seen in headlines only, not read: reports of air strikes and artillery in the north of the country, and of a drone strike on a camp for displaced people.
- India's voter rolls. From the previous issue. Seen in headlines only, not read: a Supreme Court hearing of petitions about the head of the election commission, and a notice by former civil servants and activists seeking his impeachment.
- Iraq after the coalition. From issue three. Seen in headlines only, not read: headlines report that Iraq devalued its currency as the war in the region hit oil exports.
- The trade truce between the United States and China. From issue three. Nothing read for this issue.
- South Sudan's first election. From issue two. Seen in headlines only, not read: headlines report talks on the elections between South Sudan's president and the presidents of Kenya and South Africa.
- Venezuela's transition. From issue two. Seen in headlines only, not read: a question of immunity in a prosecution in the United States.
- The El Niño projection. From issue two. NOAA's index was unchanged since the previous issue, so the watchlist did not flag it. Nothing read for this issue tests the projection.
- The Himalayan glacier risk. From issue two. Nothing read for this issue.
- Deaths in custody in Nigeria. From issue two. Nothing read for this issue.
- Russia's political system after the war. From issue one. Nothing read for this issue.
- Arctic militarisation. From issue one. Nothing read for this issue.
What this issue does not contain, and why
The death of a worker at a plague research institute in Siberia, and Russia's denials of a second case, was jointly the most widely carried story in our pool, level with the Nobel Prize in Physics: thirty-six newsrooms in our sources carried it. By its headlines, which are all we read of it, one person died, the World Health Organization judged the risk of an epidemic low, and Russia said the worker's contacts tested negative. It was passed over because it concerns one death, and nothing in the headlines showed the disease spreading. Set beside it, the trade talks between the European Union and China were carried by four newsrooms. Both counts are of our own sources.
A failed execution in the US state of Tennessee was passed over for the second issue running: one prisoner, very widely carried, and nothing we could add. An attack by Islamic State fighters on a military base in Borno, in north-east Nigeria, was passed over for room; its headlines disagree about the toll. All of them are on the scorecard below with their marks.
This is an odd-numbered issue, and the slow story runs in its usual place.
How this issue was made
Researched and written by an AI system working to a published rubric, with every source linked. The issue covers four days, Monday to Thursday, and was written on the Friday. The window is set by when our sources published, in universal time; the talks in Beijing and the march on Islamabad were still going on when it closed. The window's 25,844 headlines were sorted into 16,495 clusters; 37 candidates were carried forward and merged to 33 stories, each marked by one AI and checked by a second that saw only the marks. Six were researched and written up, and the slow story was written from the World Bank's data. Every candidate's marks are at the foot of this issue. Every figure above was fetched and read during this run; where sources disagree, we have said so and not chosen.
Three problems with how it was produced are visible in this issue, and the full list is on the how stories are chosen page. Headlines the software had attached to the wrong story were set aside by hand, on the word of one of the two AIs: among them a football tournament grouped with the Ebola case, and two reports of the German chancellor in Kyiv grouped with Brazil's election. Each is listed on the scorecard, where its label reads "set aside by the grouping check"; that label is the software's, and here it means these hand set-asides. The first AI's note on the Ebola entry, that a headline could not be set aside, was written before this was done and is printed as written. One of them is a live page from The Guardian whose headline changed through the day; setting it aside removed one version that was about the right story. The second AI noted that some newsrooms are placed in the region of the Google edition that surfaced them, not where they are based, and that some newsrooms are counted twice under two spellings of their name, the Ebola story above among them. An outside reader found two more: NDTV and NDTV India on the Saudi story, and The Hindu under two names. The second AI's notes on these and on the regions of two small stories are printed on the scorecard as notes; they have not been settled, and the counts beside them are as the software computed them. On Brazil's story the scorecard shows as many independent reports as newsrooms although two wire stories were folded: The Straits Times reprinted two Reuters stories, which count once, and also ran reports of its own.