August 28, 2026
473 owners received the prescribed oil viscosity change and 26 received an entirely new engine, and all 499 of them came back to federal investigators with the same complaint. That figure sits at the center of Engineering Analysis EA26005, which the Office of Defects Investigation opened on August 20 against 997,743 General Motors vehicles built for the 2021 through 2026 model years with the 6.2 liter L87 V-8. The question it asks is not whether General Motors should recall these trucks. It already did.
Recall 25V-274 covered L87 engines built between March 1, 2021 and May 31, 2024, and General Motors attributed the root cause to a supplier manufacturing and quality problem. The remedy branched two ways after an inspection: a change of oil viscosity, or a replacement engine. Investigators opened a recall query in January to test whether that remedy worked. Seven months of complaints answered the question.
191 additional reports concern engines built after the production window the recall covered, which is the detail that moves this from a remedy problem to a scope problem. If the failure continues in engines assembled outside the dates General Motors identified, then the supplier explanation either understates the affected period or does not describe the whole defect. Total incidents in the agency’s file now stand at 690. General Motors separately reported receiving 6,953 complaints of post-remedy engine failure.
An Engineering Analysis Is the Last Room Before a Finding
One crash and one injury appear in the federal file, which is a small number against a population approaching a million Chevrolet Silverado, Tahoe, Suburban, GMC Sierra, Yukon and Cadillac Escalade units. Loss of motive power is not primarily a crash risk. It is a stranding risk, a warranty risk and, for anyone running a fleet on these platforms, a duty cycle risk.
EA26005 is an engineering analysis, which is the final investigative stage before regulators can make an initial decision that a defect exists and press for a recall. Opening one against a remedy that has already been executed at scale is the agency saying, on the record, that it is not satisfied the problem was solved. General Motors deserves credit for issuing the original recall and for replacing engines rather than only prescribing an oil change. That credit does not answer the file.
The Second File Crossed a Corporate Boundary
August 21 is when the agency opened a separate analysis into the eBoost brake system, estimated at 1,164,820 vehicles across twelve General Motors models from the 2023 through 2026 model years. The file records 745 unique failure reports, 22 crashes or fires and six injuries. That file began as a preliminary evaluation opened April 11, 2024 on complaints about the 2023 Cadillac Lyriq. General Motors has described an internal spindle that can fracture, costing brake assist, stability control and traction control once the vehicle stops, with speed then limited to 43 mph. Newer complaints describe assist disappearing mid-stop instead, which is the inconsistency that drove the escalation.
The Honda Prologue and Acura ZDX are in that population alongside the Cadillac Lyriq, because General Motors builds them under its joint venture with Honda. A braking subsystem sourced by one automaker is now the subject of a federal investigation touching another automaker’s showrooms, its service lanes and its owner satisfaction scores. Honda has no engineering authority over the part and no control over the remedy schedule, and it will still take the customer calls.
Fewer Campaigns, Much Larger Ones
23.48 million US vehicles were recalled in the first half of 2026 against 11.09 million a year earlier, a rise of 111.6%, while the number of automotive recall campaigns fell 6.5% to 431, the lowest count in six years. Those figures come from Sedgwick’s recall index, and the firm attributes the divergence to shared platforms, shared electronics and shared software across model lines. Two files opened in two days against one automaker are what that attribution looks like in practice.
Dealer groups, insurers, lessors and used vehicle buyers that model defect exposure nameplate by nameplate are now modeling the wrong unit. Exposure follows the engine family, the brake module and the software build, and those cross nameplates, brands and in the eBoost case corporate ownership. Concentration risk on a high volume architecture is a different quantity from concentration risk on a high volume model.
Neither analysis obliges General Motors to do anything yet. What both do is date the question. On the L87 file the company has to defend a remedy it has already delivered to hundreds of thousands of owners, which is harder than defending a design, because the defense has to explain the repeat visits. On the eBoost file it has to answer for a subsystem it also sold into another automaker’s product line. The engine question is whether the fix was ever right. The brake question is how far a single General Motors part reaches. The recall arithmetic says the answer to the second one keeps getting larger.
