EPA found no unreasonable cancer risk from using laundry detergents like Tide.

1. Pass Title matches the evidence

   Title that EPA found no unreasonable cancer risk from using laundry detergents matches the EPA evidence item and the False finding.

2. Pass No obvious contradictions

   Finding False, answer, and conclusion agree; the drinking-water risk is stated consistently across conclusion, evidence and key points.

3. Pass Plausible justification

   Conclusion explains why the rating is False (affirmative EPA exposure findings for consumer use), while noting the moderate confidence and open litigation.

4. Pass Sources are relevant

   Sources (IARC, NTP, EPA, Holland & Knight, Newsday/CCE, NY DEC, NY Senate, As You Sow, CA AG, Quint, Beveridge & Diamond, EDF) are on topic by name and description.

5. Pass Common questions answered

   Five common questions are present and each response plausibly addresses its question.

Coverage: Read the full article payload; sources judged by name/title/description/role only, per First Look's rules; no source snapshots read.

1. Concern (minor) Context is complete

   Note that EPA itself characterizes 1,4-dioxane as 'likely' carcinogenic (as the cited Newsday piece says) and stop grouping NTP's 'reasonably anticipated' listing under 'possible'.

   The article is strong on context: it separates hazard from risk, notes no unreasonable risk is not zero risk, flags the lack of aggregate exposure assessment (citing EDF), the drinking-water pathway, location dependence, the pending reconsideration, and the overdue rule. One gap: the hazard evidence is titled 'Two cancer agencies classify 1,4-dioxane as a possible human carcinogen', and a prior changelog entry says the classifications call it 'a possible, not likely, carcinogen'. But NTP's category ('reasonably anticipated to be a human carcinogen') is not IARC's 'possible' tier, and the article's own cited Newsday source says 1,4-dioxane is 'Deemed a likely human carcinogen by federal agencies'; The Quint calls it 'a probable carcinogen'. The article does not address EPA's own hazard descriptor (from my own knowledge, EPA's IRIS assessment uses 'likely to be carcinogenic to humans'; I could not verify this from the cited sources). The conclusion body states each classification correctly, and the central consumer-risk finding does not depend on the hazard tier, so a reader's belief about using Tide is unchanged; hence Minor. Confidence: Medium.

   About evidence:hazard-classification, conclusion, source:cce-2024-testing, source:ntp-roc The finding

2. Pass Central claims are accurate

   Summary: the article says ordinary use of Tide does not cause cancer (rated False, Moderate confidence) because EPA's exposure-based TSCA reviews found no unreasonable risk to consumers, while noting 1,4-dioxane is an animal carcinogen, that EPA found a drinking-water risk from products washed down the drain, and that levels fell under a 2013 settlement and New York's limits. Claim ratings: (1) 2020 EPA consumer finding - True: EPA page: risks to consumers from 'eight conditions of use involving ... detergents and cleaning products' 'did not present an unreasonable risk to consumers or bystanders.' (2) 2024 revision did not identify consumer-use risk but found drain-to-drinking-water contribution - True, matches EPA page nearly verbatim; H&K's listing of 'laundry and dishwashing products' among contributing conditions of use is consistent with that pathway. (3) EPA would begin a TSCA s.6 rule; proposal due Nov 2025 - True (EPA page; H&K 'by Nov. 14, 2025'). (4) IARC Group 2B, inadequate human / sufficient animal evidence - True (IARC 5.5). (5) NTP 'reasonably anticipated' and the rodent/guinea-pig tumours - Unverifiable from NTP PDF, but animal findings Mostly True via IARC 5.3. (6) Settlement: below 25 ppm by Sept 30, 2013, case CGC-12-517674, judgment Jan 2013 - True (CA AG record; note the record says modifications 'targeted for completion no later than September 30, 2013', slightly softer than 'required'). (7) 2019 6.1-14 ppm and Jan 2024 <1 ppm for five named products; P&G compliance statement - True (Newsday). (8) NY 2 ppm / 1 ppm limits and waivers ending Dec 30, 2025 - True (DEC). (9) Union Carbide v. EPA No. 24-60615, stay to March 30, 2026, EPA's May 2025 12-24 month reconsideration - True (B&D). (10) No Europe/80-country ban - True (The Quint). (11) Finding 'False' - Mostly True: supported by EPA's consumer determination and the article's own caveat that it is not zero risk. Peripheral: evidence title calling both classifications 'possible' is Mostly True (addressed under context-and-omissions). No central claim is Mixed or worse. Confidence: High.

3. Pass Claims are supported and labeled

   Each central claim maps to cited evidence: the consumer-use finding and drain pathway to epa-final-risk-evaluation (primary) and hk-summary; hazard to iarc-vol71/ntp-roc; reductions to cce-2024-testing, ca-ag-settlement, asyousow-settlement, ny-dec-limits; regulatory status to bd-tsca-2026. The article separates fact from inference: it calls 'no unreasonable risk' 'a regulatory judgement that the estimated risk is acceptable, not a finding of zero added risk', describes the drop in levels as one that 'indicates' (not proves) reformulation, and labels P&G's compliance statement 'the company's own claim'. The only unsourced peripheral detail noticed is 'Union Carbide, a Dow subsidiary', which is a nitpick. Confidence: High.

4. Pass Sourcing is strong and transparent

   The central risk claim rests on EPA's own page (primary), with IARC and NTP primary for hazard, the California AG's court record for the settlement, and NY DEC guidance plus bill text for the limits. Weaker sources are used for peripheral points and are labelled: Newsday report of advocacy-commissioned testing ('commissioned by an advocacy group', laboratory not named), law-firm summaries for litigation ('Our account of the litigation relies on a law-firm summary rather than court records'), EDF as 'Advocacy critique', and As You Sow as 'Litigant's account'. Source-to-claim mapping via sourceIds is clear. Two sources (ntp-roc, ny-s4389b) could not be read by the server but their claims are corroborated by other cited sources. Confidence: High.

5. Pass Framing is fair and neutral

   The original claim is stated plainly ('Using Tide laundry detergent causes cancer') and the article presents the strongest version of the other side, conceding in an objection that 'EPA itself says laundry detergent contributes to cancer risk' is 'accurate, and it is the strongest version of the claim'. It labels advocacy groups and P&G's statement alike as interested parties, applies the same skepticism to industry compliance claims and advocacy testing, and avoids loaded language. The headline 'EPA found no unreasonable cancer risk from using laundry detergents like Tide' accurately reflects the EPA source. The hazard-tier wording is handled under context-and-omissions and is not double-counted here. Confidence: High.

6. Pass Uncertainty is stated

   Confidence is 'Moderate' and the reasons are explicit: the finding 'rests on one agency's modeling, which EPA is now reconsidering', 'no epidemiological study has directly measured cancer outcomes in detergent users', the 2024 testing 'covered a handful of products sold in New York and was not independently replicated', and 'A revised evaluation, a future rule, new testing, or new human data could change parts of this assessment.' Conflicting evidence (EDF's aggregate-exposure critique vs. EPA's use-by-use determination; consumer-use vs. drinking-water findings) is presented and weighed. A skeptic could argue a False rating is a step strong for a claim that rests on a regulatory acceptability judgment under reconsideration, but the article explains why it chose False over No evidence and pairs it with moderate confidence, so I do not count this as a finding. Confidence: Medium.

Coverage: Examined the full article payload of version 8 (title, claim, finding False, answer, Moderate confidence, summary, context, conclusion, keyPoints, all four evidence items with limitations, five objections, eight method steps, editorial note) and read 10 of 12 sources via get_source_snapshot (archived copies): iarc-vol71, epa-final-risk-evaluation, hk-summary, cce-2024-testing, ny-dec-limits, asyousow-settlement, quint-ban-check, ca-ag-settlement, bd-tsca-2026, edf-aggregate. Not readable: ntp-roc and ny-s4389b (both PDFs; the server reports it cannot read application/pdf). The claims those two support were cross-checked against other sources: the animal tumour findings attributed to NTP are corroborated by IARC Vol. 71 (liver tumours in mice, nasal-cavity tumours in rats, liver tumours in guinea pigs), and the New York limits attributed to the bill text are corroborated by the DEC page and Newsday; NTP's exact wording ('reasonably anticipated') was not verified against its own text. Did not open the linked EPA PDFs (2020 evaluation, 2024 revised determination), the scanned settlement/judgment PDFs, or the court docket; images were not examined beyond captions/alt text. No live web search; own knowledge used only to raise questions.

1. Concern (minor) Rates the real claim

   State that the False rating rests on U.S. evidence, or address whether Tide sold outside North America, which is not under New York's limits or EPA review, differs.

   The claim "Using Tide laundry detergent causes cancer" circulates in this form (plus the 'banned in Europe' and 'contains a carcinogen' variants, which the objections address). Narrowing to personal use and treating the drinking-water pathway separately is fair and is explained in the second objection, so this is not a strawman. However the claim is not place-limited and Tide is sold worldwide, while every risk and exposure line is U.S.: EPA's TSCA evaluations, New York's 1 ppm cap, the NY-bought 2024 samples, and P&G's statement that product "shipped in North America" complies. The conclusion says "We rate the claim about using Tide False" without restricting it to the U.S. market. A small scope mismatch; a reader would come away with the same belief for U.S. Tide.

   About conclusion, evidence:epa-risk-evaluation, evidence:product-testing, evidence:legal-limits The finding

2. Concern (minor) Strongest objections answered

   Add an objection on past exposure: whether EPA's modeled concentrations cover Tide before the 2013 settlement, when 1,4-dioxane levels were higher.

   The strongest objections are answered with evidence: hazard vs. exposure (objection 1), EPA's drinking-water finding (objection 2, conceded as "the strongest version of the claim" and fairly scoped), the ban rumor, labeling, and 'free & clear' products; aggregate exposure is handled in the EPA limitation. A secondary alternative is missing: the claim is not time-limited, and people who used Tide for years before the 2013 settlement required levels below 25 ppm were exposed to higher concentrations than the 2019 (6.1 to 14 ppm) or 2024 (<1 ppm) figures. The article does not say what concentrations EPA's 2020 consumer modeling assumed, so a reader cannot tell whether the 'no unreasonable risk' finding covers historical use. The article should address this; it would not likely change the rating, so Minor.

   About conclusion, evidence:epa-risk-evaluation, evidence:legal-limits The finding

3. Pass Rating follows from the evidence

   Before reading the finding I would have rated this False at Moderate confidence. Hazard evidence is animal-only at high drinking-water doses with "inadequate evidence in humans"; EPA's exposure-based reviews in 2020 and 2024 found consumer use of laundry products "did not present an unreasonable risk"; measured levels fell to below 1 ppm. The article explains why it chose False over No evidence ("EPA's exposure modeling is affirmative evidence about people doing laundry, not merely an absence of studies") and is candid that "no unreasonable risk" is not zero risk. Misleading would also be arguable because of the true kernel (an animal carcinogen is present, and down-the-drain use contributes to drinking-water risk), but the article rates personal use and handles the kernel explicitly, so False follows. The answer is consistent with the finding.

4. Pass Confidence matches the evidence

   Confidence is Moderate, and the conclusion gives the right reasons: the key finding "rests on one agency's modeling, which EPA is now reconsidering", "no epidemiological study has directly measured cancer outcomes in detergent users", and the drinking-water finding. Limitations cover single-agency dependence, lack of aggregate-exposure assessment (with the EDF critique), advocacy sponsorship and non-replication of testing, and pending litigation/rulemaking. The title and summary ("EPA found no unreasonable cancer risk from using laundry detergents like Tide") attribute the finding to EPA and do not overstate it; the summary also mentions the animal carcinogen and drinking-water link. I would give the same confidence.

5. Pass Evidence is independent

   The article itself flags the main dependency: IARC and NTP "both rest on the same body of rodent drinking-water studies, so they are two reviews of shared evidence rather than two independent lines." The EPA line cites EPA plus two law-firm summaries and an EDF critique, all derived from EPA's own documents, and the article labels it "a single agency's modeling". The 2019 and 2024 test figures used in both product-testing and legal-limits come from the same Citizens Campaign testing, which is disclosed. The decisive conclusion is not presented as resting on more independent lines than it has, and confidence is set accordingly, so no independence problem is hidden.

Coverage: Examined the full article payload of version 8: claim, finding (False), answer, confidence (Moderate), title, summary, key points, context, editorial note, conclusion, all four evidence items with methods, institutions, years and limitations, the five objections and responses, the eight method steps, and the source list (names, titles, descriptions). Per the Skeptic prompt, source snapshots were not read and the live web was not used; whether sources say what the article claims was not assessed (Source Checker's remit).

1. Concern (minor) Citations support the evidence

   Limitation epa-risk-evaluation's 'Union Carbide, a Dow subsidiary' and 'smaller or slow-moving surface waters' appear in no readable cited source; cite a source or cut them.

   All archived copies are the documents cited (titles, authors, dates match). hazard-classification: IARC archive supports 'There is inadequate evidence in humans... sufficient evidence in experimental animals... possibly carcinogenic to humans (Group 2B)' and oral tumours in mice, rats (incl. nasal cavity) and guinea pigs (liver). The NTP RoC profile (ntp-roc) is a PDF FactHub cannot read, so the NTP listing itself is NotChecked. epa-risk-evaluation: EPA page supports 2020 'did not present an unreasonable risk to consumers or bystanders', 2024 'does not identify risks to consumers from the use of these products', the drain-to-drinking-water finding, and 'EPA will begin developing a proposed rule under TSCA section 6'. Holland & Knight supports laundry/dishwashing products in 2020 and the Nov 14, 2025 deadline; Beveridge & Diamond supports Union Carbide Corp. v. EPA, No. 24-60615, stay until March 30, 2026, May 2025 declaration, 12-24 months; EDF supports the aggregate-exposure critique. Two limitation details are not in any readable cited source: Union Carbide being 'a Dow subsidiary' (B&D names only 'Union Carbide Corp.') and drinking-water risk being highest in 'smaller or slow-moving surface waters' (EPA page says only 'surface water contaminated with 1,4-dioxane that is discharged from industrial facilities' or products washed down the drain). These are peripheral, so Minor. product-testing: Newsday repost supports every figure and P&G's statement. legal-limits: CA AG record supports case CGC-12-517674, judgment 01/22/2013, 25 ppm; As You Sow supports the suit's basis and nationwide reformulation; NY DEC supports 2 ppm/1 ppm dates and waiver expiry. ny-s4389b (bill text PDF) could not be read, so the statutory citation is NotChecked, though DEC's page independently supports the limits.

   About limitation:epa-risk-evaluation The finding

2. Concern (minor) Quotes and figures match

   CA AG record says the 25 ppm changes were 'targeted for completion no later than September 30, 2013'; the article says the settlement 'required' it by that date.

   Checked figures: 6.1-14 ppm (2019) and <1 ppm (January 2024 tests) match Newsday; 25 ppm, CGC-12-517674, settlement 11/30/2012 and judgment 01/22/2013 match the CA AG record; 2 ppm from Dec 31, 2022, 1 ppm from Dec 31, 2023 and 'no waiver will be effective after December 30, 2025' match NY DEC; Nov 14, 2025 deadline matches H&K; No. 24-60615, March 30, 2026, May 2025, 12-24 months match B&D; P&G quote ('all product being shipped in North America is compliant') matches. Two paraphrases are slightly stronger than the source: (1) the CA AG injunctive term says modifications 'to reduce the presence of 1,4 dioxane below a level of 25 parts per million' are 'targeted for completion no later than September 30, 2013', while the evidence says P&G 'agreed to reduce... by September 30, 2013' and the conclusion says the settlement 'required it to get 1,4-dioxane below 25 ppm... by September 2013'; (2) B&D says EPA 'intends to develop additional steps to revise the risk evaluation, depending on the outcome of the reconsideration of the scientific issues', while the limitation says EPA 'intends to revise the risk evaluation after reconsidering scientific issues', dropping the conditional. A careful reader would come to much the same belief, so Minor. NTP and S4389B figures could not be checked (PDFs unreadable).

   About evidence:legal-limits, conclusion, limitation:epa-risk-evaluation The finding

3. Concern (minor) Sources represented fairly in context

   EPA's page also says manufacture of these consumer products (ethoxylation byproduct) drives worker and fenceline risk; the article frames 2024's change as only the drain pathway.

   The consumer-use findings are accurate in context. But the same EPA paragraph adds that 'the manufacture of those consumer products... the generation of 1,4-dioxane as an ethoxylation process byproduct... does significantly contribute to the unreasonable risk determination, due to worker risks... and risk to fenceline communities', and Holland & Knight lists 'laundry and dishwashing products' among conditions of use that contributed to the 2024 unreasonable-risk determination. The article reports only the down-the-drain pathway for these products ('What changed in 2024 is the drinking-water pathway'). This omission does not change the impression about consumers using Tide, which is what the article rates, so Minor. Other sources read in full: IARC's evaluation also notes tumours at other sites (subcutaneous, mammary, peritoneal) and a small negative worker study, consistent with the article's 'inadequate human evidence'. Newsday describes 1,4-dioxane as 'a likely human carcinogen' per federal agencies, but that is not cited for any classification. EDF, As You Sow, CA AG, NY DEC and B&D contain no contrary findings the article omits. NTP and S4389B were not read (PDF).

   About evidence:epa-risk-evaluation, conclusion The finding

4. Concern (minor) Conclusion rests on cited evidence

   The conclusion's 'no epidemiological study has directly measured cancer outcomes in detergent users' is not stated in any evidence item; add it to an evidence item or limitation with a source.

   Hazard layer traces to evidence:hazard-classification; EPA 2020/2024 consumer and drinking-water findings, the not-zero-risk explanation, aggregate exposure caveat, Union Carbide stay, 12-24 month reconsideration and missing Nov 2025 rule trace to evidence:epa-risk-evaluation and its limitation; 25 ppm settlement, 6.1-14 ppm, NY 2/1 ppm limits, P&G compliance statement and January 2024 retests trace to evidence:legal-limits and evidence:product-testing. One assertion has no evidence item: 'no epidemiological study has directly measured cancer outcomes in detergent users'. The nearest support is limitation:hazard-classification ('both record inadequate human evidence'; 'Neither agency evaluated laundry detergent specifically'), which is related but narrower. It is a caveat lowering confidence rather than load-bearing support, so Minor.

   About conclusion The finding

5. Pass Sources can support their claims

   Hazard claims rest on IARC (WHO agency) and NTP (federal program), both accountable scientific assessments. EPA findings rest on EPA's own page. Litigation and deadline facts rest on two named law-firm publications (Holland & Knight, Beveridge & Diamond) with named authors; the article discloses this reliance. The EDF blog is cited for EDF's own argument ('the supplemental risk evaluation systematically fails to account for the combined... exposures'), which an advocacy source can support. The product-testing evidence rests on a Newsday article by Lisa L. Colangelo, reposted on the advocacy group's site with the original byline and URL; Newsday is an accountable news publisher. As You Sow is cited for its own account of the settlement alongside the CA AG's official record. NY DEC and the NY Senate bill text are official. The Quint (supporting only an objection) is an accountable fact-checking desk. No open-edit, satire, or fabricating sources.

6. Pass Sources are archived

   Every source cited by an evidence item (iarc-vol71, ntp-roc, epa-final-risk-evaluation, hk-summary, bd-tsca-2026, edf-aggregate, cce-2024-testing, ca-ag-settlement, asyousow-settlement, ny-dec-limits, ny-s4389b) has an archiveUrl in the article, and each snapshot's copy was 'archive'. quint-ban-check is also archived. The two archived PDFs (ntp-roc, ny-s4389b) returned HTTP 200 but could not be parsed; that limits this audit, not their archival status.

Coverage: Read the archived copy (copy: archive) of 10 of 12 sources in full: iarc-vol71, epa-final-risk-evaluation, hk-summary, cce-2024-testing, ny-dec-limits, asyousow-settlement, quint-ban-check, ca-ag-settlement, bd-tsca-2026, edf-aggregate. Could not read 2: ntp-roc (NTP Report on Carcinogens profile) and ny-s4389b (NY Senate bill S4389B), both archived PDFs that FactHub cannot parse ('application/pdf' unsupported). So NTP's listing and animal-study details and the statutory text were not checked directly; NY DEC's page independently supports the limits. quint-ban-check supports only an objection, not an evidence item. Only snapshot text was used; no web browsing or outside knowledge.

1. Concern (minor) Consistent house style

   Use American "judgment" (not "judgement") throughout, explain or reword the tool name "archive_source" in method step 8, spell out WHO, and reconcile the 2013 settlement dates.

   HOUSE-STYLE rules: American English spelling; technical terms and abbreviations explained at first use; dates and names used the same way throughout. (1) Spelling: "a regulatory judgement" appears in the conclusion, the epa-risk-evaluation summary, and the third key point, while the ca-ag-settlement description uses "judgment dated January 22, 2013"; American English is "judgment", and the article is inconsistent. (2) Unexplained technical term: method step 8, "Ran archive_source on each cited URL and recorded the returned digests and any archive links," uses an internal tool name and the term "digests" a general reader would not know; suggest "Saved an archived copy of each cited web page and recorded its archive link and content fingerprint." (3) Unexpanded abbreviation: the iarc-vol71 source description says "WHO cancer agency's evaluation" without spelling out World Health Organization. (4) Inconsistent dates for the same event: the asyousow-settlement description refers to "the January 24, 2013 settlement," while the ca-ag-settlement description gives "settlement dated November 30, 2012, judgment dated January 22, 2013"; align the wording (e.g. "the settlement entered as a judgment in January 2013") or attribute the differing date. Minor names inconsistency also: "Fifth Circuit" in evidence vs. "5th Cir." in the bd-tsca-2026 description. Each is a lapse a careful editor would fix but a reader would barely notice; no leftover drafting text or conflicting figures were found (ppm figures 25, 6.1 to 14, 2 and 1 are consistent throughout).

   About conclusion, evidence:epa-risk-evaluation, method:8, source:iarc-vol71, source:asyousow-settlement, source:ca-ag-settlement The finding

2. Pass Title states the finding

   Title "EPA found no unreasonable cancer risk from using laundry detergents like Tide." states the finding in positive form, names the claim's subject (using Tide), and resolves the claim for a title-only reader without repeating the myth or using loaded words (TITLE-FORM rule: lead with what is true). seoTitle "Does Tide detergent cause cancer? EPA found no unreasonable risk from using it" uses search wording and still states the finding, as the rule permits.

3. Pass Verdict fields agree and stand alone

   Read as: FactHub rates the claim "Using Tide laundry detergent causes cancer." False (Moderate confidence): "EPA found no unreasonable risk to people using laundry detergents; its 1,4-dioxane concern is contaminated drinking water." The claim is worded as it circulates, is not its correction, and reads naturally in quotes; the answer stands alone and says what is true. Title, answer, summary ("found no unreasonable risk to people using laundry detergents") and conclusion ("We rate the claim about using Tide False rather than No evidence") all express the same conclusion in the current label's terms; no stale "No evidence" or "Mostly false" wording remains outside the changelog.

4. Pass Each section does its job

   Summary is one sentence naming the strongest lines (EPA's 2020 and 2024 exposure-based reviews and product tests). keyPoints state what the evidence establishes. Context is one paragraph of essential framing (what 1,4-dioxane is, hazard vs. risk, scope caveat on pods). Conclusion synthesizes the hazard and risk layers, explains why False rather than No evidence, justifies Moderate confidence, and lists what remains open ("What remains open: the 2024 testing covered a handful of products..."), so it does not merely restate the verdict. editorialNote states scope, caveats, regulatory currency, and "No external specialist review has been recorded." No hand-written independence note or evidence-map summary appears.

5. Pass Limitations are specific

   Each limitation is specific to its item: hazard-classification notes hazard vs. exposure, shared rodent studies, high doses, and the 1999 date; epa-risk-evaluation notes single-agency modeling, no aggregate exposure, location-dependent drinking-water risk, pending litigation, and reliance on a law-firm summary; product-testing notes advocacy sponsorship, unnamed lab, few New York products, P&G's self-reported compliance, and no replication; legal-limits notes that limits are policy choices and that the signed judgment could not be searched. None is boilerplate (LIMITATIONS-SUBSTANTIVE rule).

6. Pass Tone is measured and respectful

   Language is measured: "indicates that manufacturers reduced the byproduct", "Confidence is moderate because...", "not a finding of zero added risk". Objection responses are short, evidence-based and respectful, e.g. "That is accurate, and it is the strongest version of the claim." The flat statements "There is no such ban" and "Viral claims that Tide is banned in Europe or 80 countries are false" are backed in-article by the cited fact-check, so they are warranted absolutes. No ridicule, sarcasm, or dismissal of people who hold the claim (MEASURED-TONE rule).

7. Pass Readers can verify it

   All twelve sources carry meaningful role labels (e.g. "Regulatory risk assessment (primary)", "Legal analysis (secondary)", "Advocacy critique"), full titles, and descriptions of content or why chosen (e.g. IARC "Chosen as the non-U.S. classification"; NTP "Independent of EPA's TSCA program"). Key claims have primary or identifiable anchors: EPA's own evaluation page, IARC Volume 71, the NTP profile, bill S4389B, the court case number CGC-12-517674, and docket No. 24-60615 for the litigation (whose reliance on a law-firm summary is disclosed). Method steps cover claim definition (step 1), evidence that does not rest on one source (step 2), sources reviewed (step 3), and recorded limits (step 7) (VERIFICATION-AIDS rule).

Coverage: Examined the full article payload of version 8: title, seoTitle, socialTitle, description, claim, finding, confidence, answer, summary, all seven keyPoints, context, conclusion, all four evidence items with summaries and limitations, all five objections and responses, all eight method steps, editorialNote, image captions and alt text, and all twelve sources' roles, titles, identifiers and descriptions. Metadata (publishedAt, reviewedAt, changelog dates, URLs, archive links, digests, sourceType) was not style-checked, per the auditor's scope. Only the article was used; no live web or source snapshots were consulted, since source accuracy is out of scope.