Corrections

When we get something substantively wrong, we fix it and log the fix here; small copyedits live in each article's revision history instead. Found an error? Use the "Report issue" button on the article (no account needed). So far we have received 2 reports and reviewed 1, taking a median of 4.8 days. 1 is waiting, the oldest since September 26, 2026. That is our record rather than a promise. See the verifiability policy for how checking works.

321 articles were corrected in the last 30 days. Every correction note and resolved report is listed below, newest first. A fix applied to several articles with the same note is shown once, listing each article; "See the change" opens that version in the article's revision history.

  1. editorial

    ClusterMAX is described as four medallion tiers plus a Not Recommended band split into Underperforming and Unavailable. The article previously listed five tiers and omitted the Unavailable state.

  2. editorial

    GPT-4 parameter count now states roughly 1.8 trillion, with 1.76 trillion identified as the figure implied by 16 experts of about 111 billion parameters each, rather than presenting the two as competing numbers. Publisher name corrected to SemiAnalysis throughout.

  3. editorial

    Attribute access routes to dated sources, canonicalize documentation URL, and strengthen sourcing

  4. editorial

    Replace unsupported live-price rankings with a cutoff-bounded, independently verified cloud GPU pricing comparison

  5. editorial

    HoloLens 2's 52-degree field of view was cited to Microsoft's hardware spec page, which publishes no display field of view (its 96.1-degree figure is the visible-light tracking cameras). Re-sourced to Kipman's reveal figure via Road to VR.

  6. editorial

    The Ego4D journal extension appeared in TPAMI vol. 47 no. 11, November 2025 (pp. 9468-9509), not 2024; the 2024 in the DOI is the early-access year. CVPR page range corrected to 18973-18990 per Crossref.

  7. editorial

    Vuzix gross loss is unapplied overhead and obsolescence, not per-unit cost. Product revenue exceeded product cost of sales in FY2024 and FY2025 per the 10-K.

  8. editorial

    Independently fact-check Claude Tag through July 28, 2026

  9. editorial

    Fact-check ChatGPT, Claude, Gemini, and Grok comparison through July 28, 2026

  10. editorial

    Two statements about competitors' absence from the video arenas were wrong (Runway's Aleph 2.0 ranks sixth on the editing board, and earlier Hailuo models still rank on the boards that exclude audio), the court in the Disney suit is now named, and a timeline row was out of order

  11. editorial

    The winget package ID OpenAI documents now resolves to the Store listing titled ChatGPT Classic; the unified app is a different Store product

  12. editorial

    The tags field has changed since the 2024 record rather than being unchanged, the listing is reachable at its public URL rather than verifiably publicly listed, and the June 2024 edit timestamp is re-cited to the capture that shows it

  13. editorial

    Give the actual years TSMC and SMIC adopted 28nm rather than an approximate interval

  14. editorial

    The roughly 80 percent order-value increase applies to Chinese domestic equipment suppliers as a group against the same period of 2025, not to NAURA alone against 2024

  15. editorial

    The RMB 228.5 million figure is AMIES's committed capital as sole shareholder rather than a purchase price, and the 28nm comparison now gives the actual TSMC and SMIC adoption years

  16. editorial

    The December 2024 rule listed eight NAURA-named subsidiaries plus two Sevenstar entities rather than six, and the Xinkailai Industrial Machinery address sits inside the same industrial park as the parent

  17. editorial

    The roughly 80 percent order-value increase applies to Chinese domestic equipment suppliers as a group against 2025, not to NAURA against 2024, and the immersion DUV scanner is produced by Shanghai Aishengna to SMEE's SSA800 design rather than by Yuliangsheng

  18. editorial

    Freeze pre-cutoff product documentation, narrow record and regulatory wording, canonicalize links, and add peer-reviewed safety evidence.

  19. editorial

    SMIC's Entity List review policy was quoted from the 2020 text as though current. BIS amended the entry on 5 December 2024, adding a Footnote 5 designation and moving it to a presumption of denial for all items, with case-by-case treatment confined to 200mm production

  20. editorial

    SemiAnalysis projected the TSMC die bank would run down within roughly nine months

  21. editorial

    The generations table listed the 910B and 910C flatly as SMIC 7nm, contradicting the teardown evidence set out in the article's own body text and in three sibling articles; the 40 percent yield figure is the Financial Times reporting Huawei's overall programme yield from unnamed sources

  22. editorial

    A direct quotation attributed to SemiAnalysis was not in the source, the Ascend die-bank package split was derived rather than reported, SMIC's N+1 node was confirmed by a 2022 TechInsights teardown, and BIS amended SMIC's own Entity List entry in December 2024 to a presumption of denial for all items

  23. editorial

    Two CNAS quotations were composites rather than verbatim, the analysis was published the day after the instruments took effect rather than two days, and the second tariff window straddles the export-rule ceiling rather than sitting inside it

  24. editorial

    MetaX, Enflame, T-Head and Kunlunxin are not on the Entity List, so the claim that almost all Chinese AI chip designers are export-controlled overstated it; also corrects two entity counts, adds periods to two market-share estimates, and attributes the market-value claim

  25. editorial

    The April 2025 capital increase was Quanhong's 1.6 billion yuan ticket for 0.99 percent within a round of about 9.42 billion, the Electronics Weekly layer-count contradiction is reported rather than silently repaired, and figures drawn from the July 2026 schematic map are identified as one compiler's estimates