Al-Shabaab and a Former President's Militia Attacked the Same Somali City for the Second Time in Three Weeks. This Time, the Government Won't Say How Many It Killed.

Abstract: Fighting broke out before dawn on 3 September 2026 in Baidoa, the capital of Somalia's Southwest State, when armed fighters struck federal military positions with a series of explosions in the Al-Faras, Boonkaay and Al-Axmaar areas of the city. Somalia's Ministry of Defence said in a statement posted on X that "militias" loyal to ousted Southwest State President Abdiaziz Hassan Mohamed "Laftagareen" carried out the attack together with fighters it described as "Khawaarij," a term Somali officials use for al-Shabaab. The ministry said its forces "responded bravely" and inflicted a "heavy defeat" on the attackers, but unlike after a nearly identical attack on the same city three weeks earlier, it did not give a combatant casualty figure, saying an assessment was still under way. What is independently documented is the civilian toll: Dr Abdikafi Omar Mohamud, head of the emergency department at Bay Regional Hospital, said the facility had received 71 wounded people, including 15 women, and one dead civilian, the highest single casualty count Baidoa has recorded since federal forces seized the city from Laftagareen's administration in March. A spokesman for Laftagareen's forces claimed they had taken full control of Baidoa; the Somali military strongly denied this and said it had prevented the attackers from seizing key positions in the city. FIKIR Institute first examined this same dispute after an almost identical attack on 17 August, when the government named Laftagareen's forces and al-Shabaab as a combined attacking force while the army's own statement credited a precise figure, roughly 30 fighters killed, to a narrower account that did not mention Laftagareen at all. The 3 September attack repeats that structure with one significant difference: in the intervening three weeks, Laftagareen went on the record to deny any connection between his forces and al-Shabaab, and the government's own account of what happened this time has become less specific, not more, even as the human cost has grown.

What happened before dawn on 3 September

Somalia's Defence Ministry said the attack began with a series of explosions intended to terrorise civilians, followed by gunfire that spread into several parts of Baidoa, including the Al-Faras, Boonkaay and Al-Axmaar neighbourhoods. Residents told wire reporters they remained indoors as the fighting continued for several hours before clashes stopped in the early afternoon. The ministry's statement described its forces as having "inflicted a heavy defeat" on the attackers without providing figures, saying it was still assessing the number of fighters killed and would release details "once the assessment has been completed." That is a marked departure from how the government handled the near-identical attack three weeks earlier: after the 17 August assault, the Ministry of Defence credited the Somali National Army within hours with killing roughly 30 attackers it identified as al-Shabaab fighters, a specific figure reported the same day by Xinhua and other outlets, even though the ministry offered no evidence for the count and did not identify those killed. No equivalent figure has been offered for 3 September, despite the government describing the defeat it inflicted in equally emphatic terms.

What is independently documented from 3 September is the civilian cost. Dr Abdikafi Omar Mohamud, head of the emergency department at Bay Regional Hospital in Baidoa, told reporters his staff had received 71 wounded people, including 15 women, along with one dead civilian. That single figure, on its own, is higher than the combined ten dead reported by residents and a local elder after the 17 August attack, and it makes 3 September the heaviest single toll Baidoa has recorded since federal forces took full control of the city from Laftagareen's administration on 30 March. The government's own account of that toll, however, has not extended past the civilian numbers reported by the hospital; the military side of the ledger remains, in the ministry's own words, still being assessed.

Two claims of control, one flatly denying the other

As after the August attack, the two sides did not agree on the outcome. A spokesman for Laftagareen's forces claimed they had seized full control of Baidoa during the fighting. Somalia's military rejected the claim outright, describing the attackers, in a formulation the government has increasingly used for al-Shabaab, as "Khawaarij," and stating that government forces had prevented them from seizing key positions in the city. Both claims cannot be true in full: either federal forces held the city, as Mogadishu says, or an opposition-aligned force briefly controlled a city of well over a million people, including more than half a million displaced by conflict and drought, as Laftagareen's spokesman says. FIKIR Institute has not identified independent verification of either claim; residents who spoke to AFP after the fighting described the city as calm following roughly six hours of clashes, which is compatible with either side's account of who ultimately held ground once the shooting stopped.

The denial that came between the two attacks

The most significant development since FIKIR Institute's 18 August report on the first attack did not happen on a battlefield. Around 28 August, Laftagareen issued a public statement rejecting the federal government's characterisation of his Darawish forces as linked to al-Shabaab, calling the allegation baseless and warning that unsupported claims of this kind put both civilians and security personnel at risk. He said the Darawish were legally established security units tasked with protecting civilians, government institutions, public security and constitutional order in Southwest State, and insisted the August operation had nothing to do with al-Shabaab. That denial was specific, public, and made roughly a week before the second attack occurred. It did not change the government's framing: the Ministry of Defence's 3 September statement again named Laftagareen's militias alongside "Khawaarij" fighters as a combined attacking force, using language nearly identical to its account of 17 August, without addressing Laftagareen's intervening denial or offering new evidence for the linkage it continues to assert.

A pattern now three months old

The 3 September attack is not an isolated flashpoint; it is the latest stage in a dispute that has run since Southwest State's parliament re-elected Laftagareen to a new five-year term in late March, a vote the federal government rejected as unconstitutional. Federal troops entered Baidoa on 30 March, Laftagareen resigned and fled to Nairobi, and Southwest's finance minister was installed as acting president. Laftagareen attempted to retake the state in May and failed. Ahead of a fresh vote to install a permanent replacement, authorities imposed security restrictions, including limits on heavy weapons, after forces loyal to Laftagareen again attacked Baidoa. Sheikh Aden Mohamed Nur "Madobe," at the time still serving as Speaker of Somalia's federal House of the People, was elected Southwest State president on 11 June, winning 88 of 90 votes cast in the state assembly after his main rival, Sharif Hassan, withdrew and dismissed the process as a "one-man election." Laftagareen's loyalists have been aligned with the Somalia Future Council, the national opposition coalition contesting President Hassan Sheikh Mohamud's own mandate, since March, tying the Baidoa fighting directly to the wider national dispute over federal legitimacy that FIKIR Institute covered in its 3 September report on Mohamud's election deal with the opposition. Two months after Madobe's contested installation, on 17 August, came the first major coordinated attack the government attributed to Laftagareen's forces and al-Shabaab together; three weeks later, on 3 September, came the second.

Why the transparency gap matters

A government that names a precise casualty figure within hours after one attack and offers none after a materially worse one, three weeks later, on the same city and against the same alleged combination of forces, is not simply being cautious. Either the August figure was reliable enough to publish immediately and the September figure is taking unusually long to verify for reasons the ministry has not explained, or the August figure was published for its rhetorical value, to demonstrate a decisive battlefield win against al-Shabaab, and the September ministry has less confidence in a number it would be willing to defend. Both explanations matter for how much weight outside observers should put on the government's account of who attacked Baidoa and what happened to them, particularly given that the underlying allegation, that a domestic political rival's forces are operationally fighting alongside al-Qaeda's regional affiliate, is one the government has repeated without new evidence after the rival specifically and publicly denied it.

What to watch

Three signals will indicate whether 3 September marks an entrenched pattern or a peak in an escalation that recedes. First, whether the Ministry of Defence publishes the combatant casualty assessment it says is under way, and whether that figure, if it appears, is accompanied by any evidence, identification of fighters killed, captured weapons, prisoners, beyond the figure itself; a long delay or an unsupported number released without corroboration would suggest the August number was similarly unverifiable. Second, whether Laftagareen issues a direct response to the 3 September attack matching the specificity of his 28 August denial, since a repeated, detailed denial after a second attack would sharpen the contradiction between his public position and the government's continued attribution. Third, whether the interval between attacks on Baidoa continues to shrink, from roughly five months between the March takeover and the August attack to three weeks between August and September, a shortening cycle that would suggest the attacking coalition, whatever its actual composition, retains both the intent and the capacity to strike the city on a recurring basis.

Policy implications

For Somalia's federal government, continuing to name Laftagareen's forces alongside al-Shabaab in official statements while declining to substantiate either that linkage or the casualty figures used to describe the government's response risks the same credibility cost FIKIR Institute flagged after the August attack, now compounded by the fact that the rival named in the accusation has since denied it on the record without a substantive government reply. For Southwest State under President Madobe, a second major attack on the state capital within three weeks of the first is evidence that the political settlement installing him has not resolved, and may be actively provoking, an armed contest for the city that neither a 88-vote assembly majority nor a federal troop presence has closed off. For international partners funding Somalia's security forces, a government asking for continued support to fight al-Shabaab while offering less verifiable detail about a worsening attack than it offered about a smaller one three weeks earlier is a reason to press for the combatant casualty assessment, and the evidence behind it, before treating either the August or the September figures as a reliable measure of operational success. For Baidoa's residents, including the more than half a million people displaced into the city by conflict and drought, the least contested fact remains the one FIKIR Institute noted in August: a city of this size, still bordered by an unresolved political-military dispute and a persistent al-Shabaab presence a few kilometres beyond its edges, does not stop being exposed to attack because the state managing it changes hands.