Home › Corrections
Corrections log
Material corrections to CoStudy question banks, most recent first. A catalogue this size will contain errors; what matters is how fast they are fixed once found.
This log records substantive corrections — a wrong rule, a stale exam blueprint, a defect that affected how a bank was scored or weighted. Routine additions and rewrites are not listed. If you have found something wrong, tell us and it will appear here once fixed.
| Date | Bank | Correction |
|---|---|---|
| 2026-10-09 | CPA Auditing and Attestation (AUD) | Sixteen items, mostly on SSARS engagement types, single audits, special purpose frameworks and network-firm independence, were keyed to a wrong answer even where the explanation described the right one. Each key was corrected and its explanation rewritten. |
| 2026-10-09 | Catalogue-wide | An answer-key audit found 217 explanations, across Series 7, Series 65, AWS, HSPT, SSAT and other banks, still labelled with the letters from before the answer choices were reordered, so the explanation pointed at the wrong choice; the keys themselves were right. The labels were corrected. The same audit fixed a wrong key or a question with no correct choice in HSPT, Grade 4–8 math, AP Microeconomics, AP Physics C, AP Chemistry and A Midsummer Night’s Dream items, and the catalogue is now scanned for the defect before content ships. |
| 2026-10-07 | Series 7 | An item on MSRB Rule G-37 (political contributions) had the wrong answer, and 280 items were duplicated through a stale copy of the bank. The key was corrected and the duplicates removed. |
| 2026-08-10 | NMLS SAFE MLO (National Test) | The bank encoded a superseded content outline — Origination Activities at 21.9%, Ethics at 14.9%, Uniform State Content at 18.8%. The current NMLS outline weights these 27%, 18% and 11%. The bank was redistributed to the current weighting and the practice exams re-split to match. |
| 2026-08-10 | NMLS SAFE MLO (National Test) | One item stated the National Test is 125 questions with 10 unscored pretest items. It is 120 questions with 5 unscored. The item was rewritten. |
| 2026-08-10 | Catalogue-wide | A length-bias audit found banks where the correct answer was the longest choice in up to 97% of items, letting a candidate score well without reading the question. Affected items were rewritten to length parity and their rationales rebuilt to address every option. |
| 2026-08-10 | Databricks Generative AI | Twenty-eight items contained a corrupted token introduced by an earlier bulk find-and-replace. Corrected, and the whole catalogue re-scanned for the same defect. |
| 2026-08-08 | CompTIA A+ | Retargeted from the retired 1100 series to the current 220-1201 and 220-1202 exams, with the domain weightings updated to the 1200-series objectives. |
| 2026-08-08 | Microsoft Azure AI | AI-900 and AI-102 retired on 30 June 2026. The banks were retargeted to their successors, AI-901 and AI-103. |