Index update, Q1 2027: fourteen cards added, four fields added, four rankings added, three log lines

The fourth wave added ten cards of four kinds. Six are text detectors: Smodin, with the highest stated language count in the index at 100+ (smodin.io, read 9 November 2026); BrandWell, tested as Content at Scale in two 2023 records and now a Moxby browser Mod; Surfer, Compilatio and Quetext; and TruthScan, which prices results rather than words. DetectGPT is the second open research method on the index, with records from Weber-Wulff 2023, Krishna 2023 and Sadasivan 2023 (open-source view). Sightengine, Deepware and Sensity are media detectors for images, video and audio with no text line, which places them on the content type matrix and in the zero-record rows of studies by content type.

The fifth wave added four cards that stretch the index into code and voice. Codeleaks is Copyleaks' source-code product, whose AI verdict was removed from the API submit endpoints on 16 December 2025 per the release notes (copyleaks.com/codeleaks, read 9 November 2026); Illuminarty is the third commercial image detector Ha and Passananti 2024 scored, at a 67.40% false positive rate on 280 artworks; Pindrop Pulse and Resemble Detect are the first two voice-first cards, both with vendor accuracy figures and no study record (pindrop.com; resemble.ai, read 9 November 2026). The sixth wave added no card; it added records and fields, below.

Four, all on 9 November 2026, each written by a merge script into the rank_fields block of every card with the same keys and a null form where no figure qualifies. At version 2026.11.5: integration_count (how many of eight named surfaces the vendor states: a browser extension, Google Docs, Word, an LMS, Zapier, MCP, a REST API, a mobile app) and file_types (distinct upload formats the input limits text names, as the vendor spells them). At version 2026.11.7: minimum_input (the smallest text length the vendor states as a requirement for a check, normalised to characters at 5.5 characters a word for ordering only, with the vendor's wording beside it) and api_price_per_1k_words (the published price of the text API per 1,000 words in US dollars, converted from the vendor's unit and shown both ways). Each field feeds one ranking: platform integrations, file types, minimum input and API pricing. The methodology page gained a row per ranking with its field, direction, qualification and tie rule, and the metrics table in the data file gained a direction for each.

Three study-fed fields also moved in the sixth wave under rules the rankings already stated. GPTZero's robustness moved from 20% (Orenstrakh 2023) to 60.04%, the true positive rate on humanized essays at a fixed 5% false positive rate in DAMAGE 2025, under the rule "highest printed post-attack detection" (arXiv 2501.03437, read 9 November 2026). Pangram's published_fnr moved from 1.31% (Jabarian and Imas 2025) to 0%, 0 of 297 plain-prompt AI passages missed in Epoch AI 2026, a printed count under the no-complement rule (epoch.ai, read 9 November 2026). The GPT-2 Output Detector's published_fnr moved from 33.3% (Weber-Wulff 2023) to 8.3%, 10 of 120 ChatGPT essays above a 5% originality cut in Ibrahim 2023 (DOI 10.1186/s40468-023-00260-2, read 9 November 2026). Hive and Sightengine gained image robustness values of 44% and 0% from RAID image 2025, labelled image. No published_fpr value moved: Epoch AI's 0 of 495 for Pangram and GPTZero is not lower than the zeros already on the cards, so it was added to the base text under the "lowest study-sourced rate" rule.

Four rankings were added, two per field pair: platform integrations (38 of 42 qualify, Walter Writes first at 6 of 8 surfaces), file types (20 of 42, Sensity first at 22 formats), minimum input (10 of 42 state a requirement, Phrasly first at 30 words) and API pricing (10 of 42 print a price and a unit, AI or Not lowest at $5 per million words), by the computed order at version 2026.11.13. Of the twelve rankings that existed at the Q4 update, independently tested went from 14 of 28 to 21 of 42, with DetectGPT, BrandWell, Compilatio, Illuminarty, QuillBot, Undetectable AI and Sightengine joining and GPTZero rising from 11 records to 16. False positive rate went from 12 to 15 ranked, three zeros from Weber-Wulff 2023 joining. Languages went from 13 to 21, Smodin's 100+ taking first place from Undetectable AI's 50+. Input limits went from 8 to 15, AI or Not first at 500,000 characters per API request. Data retention went from 9 to 13. Studies by content type went from 16 to 24 and from one record on a non-text type to two, Hive now holding both. Adversarial robustness went from 8 to 11, DetectGPT, Hive and Sightengine joining and GPTZero rising from eighth to fifth. False negative rate went from 12 to 16, Pangram rising to the zero group. Claim vs evidence gap went from 8 to 9 with Compilatio at plus 15 points, the widest gap. Free tiers stayed at 4. Every change is dated in the revision log at the foot of the ranking it touched.

Three lines in the corrections log since the Q4 update, all additions and re-sorts rather than corrections to a quoted figure: the wave 4 line (version 2026.11.5: ten cards, two rankings, four questions, nine memberships changed), the wave 5 line (version 2026.11.6: four cards, two records with AIDI-S-0020 reserved for Dalalah 2023, six comparisons, two views, one question) and the wave 6 line filed with this update (version 2026.11.7: eleven records with AIDI-S-0022 reserved for Jiang 2024, two fields, two rankings, two updates, three fields moved by rule). Two register corrections travelled with the sixth wave: the Kar 2024 record's slug and title were corrected from a placeholder, and the DFDC 2020 register note's "average precision 65.18%" was corrected to accuracy, the word Meta's results post uses (ai.meta.com, read 9 November 2026). No price, limit or study figure quoted on a card has been corrected since launch; the first scheduled recheck of prices and free tiers on all cards is 8 December 2026, and any change it produces will be its own line.

• The index data file, versions 2026.11.4 to 2026.11.7 (read 9 November 2026)
• Smodin AI detector: https://smodin.io/ai-content-detector (read 9 November 2026)
• Copyleaks Codeleaks: https://copyleaks.com/codeleaks (read 9 November 2026)
• Pindrop Pulse for Contact Centers: https://www.pindrop.com/pulse-for-contact-centers (read 9 November 2026)
• Resemble Detect: https://www.resemble.ai/products/detect (read 9 November 2026)
• Masrour, Emi and Spero 2025, DAMAGE: https://arxiv.org/abs/2501.03437 (read 9 November 2026)
• Epoch AI 2026, AI detectors rarely flag human writing: https://epoch.ai/data-insights/ai-detectors-false-negatives (read 9 November 2026)
• Ibrahim 2023: https://doi.org/10.1186/s40468-023-00260-2 (read 9 November 2026)
• Eddoubi et al. 2025, RAID image benchmark: https://arxiv.org/abs/2506.03988 (read 9 November 2026)
• Meta AI, Deepfake Detection Challenge results, 12 June 2020: https://ai.meta.com/blog/deepfake-detection-challenge-results-an-open-initiative-to-advance-ai/ (read 9 November 2026)

Revision log: 2026-11-09: update created at index version 2026.11.7.