AI Wiki

Explore AI Wiki

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

    Independently reverified GPT-3.5 family scope, API snapshots, documented specifications, fine-tuning prices, empirical limitations, and deprecation status.

  2. editorial

    Independently reverified Figure 02 specifications, Helix scope, BMW deployment evidence, successor attribution, and source quality.

  3. editorial

    Reconstructs the classification article with verified definitions, bounded algorithm history, task structures, evaluation, calibration, imbalance, uncertainty, fairness, governance, and 41 explicit sources; removes unsupported rankings, prevalence figures, and universal claims.

  4. editorial

    Independently verify Andreessen Horowitz's legal identity, fund chronology, regulatory assets, operating model, and dated media and policy activity; remove unsupported portfolio, performance, valuation, partner, and office claims

  5. editorial

    Independently verify and rebuild Batch Normalization with exact equations, batch-axis semantics, bounded mechanism evidence, small-batch and distributed caveats, alternatives, and versioned framework behavior

  6. editorial

    Independently reverified and rebuilt the ROS article; distinguished ROS 1 from ROS 2, corrected current distribution and end-of-life status, bounded middleware interoperability, real-time and security claims, and replaced unsupported adoption material with 31 explicit primary or scholarly references.

  7. editorial

    Independently verify Google Brain's organizational history and technical record, replace unsupported project ownership and causal claims, strengthen primary and peer-reviewed citations, and preserve Google Research and Google DeepMind identity boundaries

  8. editorial

    Update Mila role, fix award dates, tighten shared research attribution, and qualify AI-safety and Scientist AI claims

  9. editorial

    Independently verify Gemma chronology, architectures, parameter accounting, modalities, licensing, specialized variants, evaluation boundaries, and safety limitations through July 28, 2026

  10. editorial

    Reconstruct AI Infrastructure with source-bounded system layers; independently verify compute, networking, checkpointing, serving, facilities, energy, governance, and benchmark claims; correct the Kubernetes source target; and remove an overbroad energy-accounting sentence

  11. editorial

    Preserve the AI Chip physical-device identity, harden taxonomy and claim-level citations, verify architecture, packaging, benchmark, supply-chain, and export-control boundaries, and retain AI Accelerator as a protected distinct page

  12. editorial

    Enacted the 2026 Digital Omnibus status, fixed phased dates and legal effects, narrowed scope and transitions, removed unsupported enforcement and cost claims, and separated law, guidance, codes, standards, and scholarship

  13. editorial

    Independently verified and rebuilt the article; corrected ALVINN, DAgger, RT-1, Open X-Embodiment, Rubik's Cube, and model-comparison claims; removed unsupported history and simulator survey material; added 23 explicit primary or scholarly references.

  14. editorial

    Fixed AlexNet ensemble scope, backpropagation attribution, Royal Medal sourcing, QEPrize credit, and an unsupported name detail

  15. editorial

    Independently verifies MIT's exact institutional identity, Dartmouth-versus-MIT AI origins, Project MAC and AI Lab chronology, current organizational boundaries, 2025-2026 scale and financial context, the 2025 Quest rename, the 2026 MIT-IBM successor lab, selected research evidence, technology transfer, and documented governance accountability through July 28, 2026.

  16. editorial

    Rebuild GPT Store history, publishing, privacy, and research claims

  17. editorial

    Independently reverify transfer learning evidence boundaries, adaptation results, and source metadata

  18. editorial

    Independently reverify knowledge distillation objectives, evidence boundaries, failure modes, and source metadata

  19. editorial

    Rebuild GSM8K size, protocol, contamination, and reliability claims

  20. editorial

    Independently reverified GAN article; fixes BigGAN variant and resolution benchmark values, states theorem assumptions, replaces unsupported claims with 30 explicit primary or peer-reviewed sources, and adds bounded training, evaluation, application, and risk coverage.

  21. editorial

    Rebuild Tsinghua University around dated official records; distinguish university units from BAAI, Zhipu, and research collaborations; remove unsupported rankings, founder-share, export-control, and promotional claims.

  22. editorial

    Clarify Meta deal, contract, labor, leadership, and valuation claims

  23. editorial

    Replaced volatile model and leaderboard claims with source-bounded coverage of VLM architectures, training, evaluation, limitations, and deployment

  24. editorial

    Removed unsupported current staff, budget, lab, leadership, product-attribution, impact, and superlative claims; replaced them with dated organization records and bounded primary evidence

  25. editorial

    Distinguish Google Cloud from GCP and fix financial and Wiz figures