GarageCalc The record is in reports, not a measured failure rate
We compared nine equal calendar windows, January 1 to August 24 in each year from 2018 through 2026. The latest count exceeds the previous high in 2022 by 62 reports. Against 2025, this category grew 19% while all vehicle complaints grew about 2%.
The scope matters. NHTSA's forward collision avoidance component category includes automatic emergency braking (AEB), adaptive cruise control (ACC) and warnings. A complaint can describe unexpected braking, a failure to brake, an incorrect warning or another problem. This is not a count of phantom-braking incidents, and it does not cover every driver-assistance feature.
In the 2026 category, 98 narratives contain the stems “phantom brak” or “ghost brak”. That narrow text search is neither a validated incident classification nor a complete measure of unexpected braking. We keep it separate from the category total.
Nine years, the same January-to-August window
Every bar below covers January 1 through August 24 by NHTSA receipt date. None is a full-year total or an annualized forecast.
| Year | All vehicle complaints | FCA complaints | FCA share | FCA crash flags | FCA crash share |
|---|---|---|---|---|---|
| 2018 | 41,374 | 70 | 0.2% | 10 | 14.3% |
| 2019 | 49,588 | 70 | 0.1% | 6 | 8.6% |
| 2020 | 37,839 | 36 | 0.1% | 8 | 22.2% |
| 2021 | 33,886 | 985 | 2.9% | 105 | 10.7% |
| 2022 | 34,823 | 3,124 | 9% | 170 | 5.4% |
| 2023 | 40,684 | 2,582 | 6.3% | 244 | 9.5% |
| 2024 | 44,909 | 2,476 | 5.5% | 236 | 9.5% |
| 2025 | 48,806 | 2,683 | 5.5% | 257 | 9.6% |
| 2026 | 49,801 | 3,186 | 6.4% | 445 | 14% |
FCA means forward collision avoidance. Early years have small counts and reflect different adoption and component-label usage. For example, 2020's 22.2% crash share is eight reports out of just 36. The 2026 crash share is not the highest in this same-window series.
Crash flags rose faster than the category total
Of 3,186 category complaints in 2026, 445 have CRASH=Y, compared with 257 of 2,683 a year earlier. The share rose from 9.6% to 14%. Across all 49,801 vehicle complaints in the 2026 window, 1,997 have a crash flag, a 4% share.
This is a comparison of complaint records. It cannot establish that a system caused a crash, that each report describes a different crash, or that one equipped vehicle is more likely to crash than another. The file does not provide a matching denominator of vehicles, miles travelled or system activations. Changes in reporting behavior and the kinds of problems being reported can change the share.
Which models contribute the most reports?
The Honda Odyssey has the largest category count, with 188 complaints. Tesla Model Y follows with 165, then Hyundai Tucson with 147. Their crash flags are 2, 55 and 31, respectively. The volume leader and the largest crash-flagged count among these three are different models, which is one reason to retain both columns.
These totals combine model years. A model name is not a specific trim, hardware version or software release. They are starting points for examining reports, not a reliability ranking or an assessment of a particular recall population. Links open our all-component model pages, whose totals are intentionally broader than this table.
| Model | 2025 FCA | 2026 FCA | Crash flags | Crash share | AEB / ACC / warnings |
|---|---|---|---|---|---|
| Honda Odyssey | 138 | 188 | 2 | 1.1% | 95 / 121 / 105 |
| Tesla Model Y | 83 | 165 | 55 | 33.3% | 81 / 94 / 39 |
| Hyundai Tucson | 31 | 147 | 31 | 21.1% | 83 / 11 / 84 |
| Honda Pilot | 72 | 126 | 6 | 4.8% | 66 / 78 / 56 |
| Honda CR-V | 113 | 121 | 11 | 9.1% | 88 / 28 / 36 |
| Honda Accord | 107 | 105 | 6 | 5.7% | 67 / 38 / 29 |
| Tesla Model 3 | 63 | 76 | 22 | 28.9% | 33 / 51 / 12 |
| Ram 1500 | 55 | 56 | 2 | 3.6% | 28 / 24 / 10 |
| Acura MDX | 29 | 53 | 2 | 3.8% | 13 / 35 / 29 |
| Nissan Rogue | 43 | 48 | 7 | 14.6% | 26 / 4 / 24 |
| Ford F-150 | 35 | 44 | 8 | 18.2% | 23 / 11 / 13 |
| Chevrolet Equinox | 8 | 43 | 4 | 9.3% | 28 / 11 / 9 |
| Honda Prologue | 79 | 43 | 4 | 9.3% | 31 / 10 / 15 |
| Jeep Grand Cherokee | 36 | 41 | 3 | 7.3% | 21 / 7 / 24 |
| Honda Passport | 21 | 40 | 0 | 0% | 24 / 13 / 21 |
| Chevrolet Silverado 1500 | 22 | 36 | 2 | 5.6% | 22 / 10 / 15 |
| Volkswagen Tiguan | 7 | 34 | 1 | 2.9% | 18 / 13 / 8 |
| GMC Sierra 1500 | 19 | 33 | 0 | 0% | 14 / 16 / 10 |
| Hyundai Santa Fe | 25 | 32 | 2 | 6.3% | 14 / 14 / 16 |
| Audi Q5 | 6 | 31 | 1 | 3.2% | 16 / 16 / 13 |
Top 20 by 2026 complaint count; equal counts are sorted by raw model name. A complaint can name several subcategories, so AEB, ACC and warning counts overlap. Some records name only the parent category.
Growth leaders with a minimum reporting base
This table requires at least 25 reports in 2026 and 20 in the matching 2025 window. The thresholds avoid ranking a change from one report to two above much larger changes, but do not normalize by fleet size.
| Model | 2025 | 2026 | Change |
|---|---|---|---|
| Hyundai Tucson | 31 | 147 | +374% |
| Tesla Model Y | 83 | 165 | +99% |
| Honda Passport | 21 | 40 | +90% |
| Acura MDX | 29 | 53 | +83% |
| Honda Pilot | 72 | 126 | +75% |
| Chevrolet Silverado 1500 | 22 | 36 | +64% |
| Honda Odyssey | 138 | 188 | +36% |
| Hyundai Santa Fe | 25 | 32 | +28% |
| Ford F-150 | 35 | 44 | +26% |
| Tesla Model 3 | 63 | 76 | +21% |
What owners can do with these findings
Use the model count to decide which complaint narratives to read, then match the model year and equipment to your own vehicle. Check open recalls for the specific VIN through NHTSA's recall lookup. This analysis cannot diagnose an individual car or determine whether a repair or software update is effective.
The Car Complaint Checker covers all components. Our earlier EV and hybrid study looks at broad model trends; this study follows one component category across equal receipt-date windows.
Methodology and limitations
- Source and snapshot: the NHTSA ODI COMPLAINTS_RECEIVED_2015-2019, 2020-2024 and 2025-2026 text archives retained locally. This study uses records received through August 24, 2026; it is not a live September feed. Source file SHA-256 hashes are included in the downloadable JSON.
- Population: vehicle products (PROD_TYPE=V), receipt years 2018–2026, January 1 through August 24 in each year. Dates use LDATE, the date NHTSA received the complaint, not the incident date.
- Counting: one record per distinct ODI number. We union component labels across repeated rows, then include the record if any label begins with FORWARD COLLISION AVOIDANCE. Duplicate IDs are not independent component complaints. Separate ODI numbers may still describe the same event.
- Crash flags: a record counts as crash-flagged if a retained row has CRASH=Y. The flag records reported crash involvement, not confirmed cause. We do not infer an injury count, a system failure rate or a per-vehicle crash probability.
- Comparability: using the same calendar window addresses unequal observation periods. It does not control for changing fleet size, vehicle age, mileage, feature adoption, category coding, publicity or owners' propensity to file a complaint. Early years warrant particular caution.
- Text mentions: the “term” field matches “phantom brak” or “ghost brak”, case-insensitively, in any row of an included complaint. It is a keyword flag, not a validated phantom-braking event count. Complaints filed only under other components are outside this study.
- Model tables: model names follow the source, with display names from our existing lookup. The download includes all 78 models with at least ten category complaints in 2026. Growth is blank if the 2025 base is below 20; absence from a table is not evidence of zero problems.
- Reproducibility: the build checks receipt-date and model consistency for repeated IDs, and checks that full model and make totals equal the category total before applying display thresholds. The analysis source is downloadable; provide the three raw archives to rerun it.
Field definitions and the role of complaints in investigations are documented in NHTSA's datasets guide and complaint file instructions. NHTSA's records are the source; the analysis and interpretation here are GarageCalc's.
Download the data
- Annual same-window counts and crash flags (CSV)
- All 78 qualifying models (CSV)
- Full aggregate output and source-file hashes (JSON)
- Reproduction script (Node.js)
CSV fields: all and fca are distinct complaint counts; allCrash, fcaCrash and crash2026 are crash-flagged counts; shares and growth are percentages. aeb, acc and warnings are overlapping component counts. term is the narrow narrative keyword flag described above. All annual rows use January 1 through August 24.
GarageCalc's aggregate tables and analysis script are available under CC BY 4.0. Please cite this study and its August 24 data cutoff. Corrections and data questions: [email protected]. See our sources and corrections policy.