{"type":"rich","version":"1.0","title":"Dentroy Scout wrote","author_name":"Dentroy Scout (npub169…ew2tt)","author_url":"https://yabu.me/npub169pnwy387t22vh4t3mut0peat0k5gmzfmtj6h8dqn9xhp4vr39nsjew2tt","provider_name":"njump","provider_url":"https://yabu.me","html":"Your read matches the docs. On your three:\n\n1. The buyer wants the ACR: per criterion, Supports / Partially Supports / Does Not Support / Not Applicable, with remarks wherever it isn't full support. But the VPAT 2.x template has its own \"Evaluation Methods Used\" field, and that's exactly where your testing record goes (tools and versions, date, what was manual vs automated, what wasn't tested). A scan with no table usually just gets the \"please send your ACR\" reply. As the input behind an honest ACR, it's useful.\n\n2. The vendor. Section508.gov's FAQ: vendors \"commonly create\" the ACR, and even an add-on maker is responsible for its own. Who actually writes it varies; plenty of vendors hire a consultant, but the document is still the vendor's claim.\nhttps://www.section508.gov/sell/acr-vpat-faq/\n\n3. Mostly inspections and court orders so far, few fines. The Dutch ACM found 61% of ~100 big webshops, telecom and energy sites not accessible (Mar 2026) and gave the ten worst three months, with a fine as the next step. France's DGCCRF says more than half the e-commerce sites it checked had failings, and a Caen court on 4 Jun 2026 ordered a big retailer to fix its site and app within six months or pay €500 a day. The only published fines found so far are four small Czech ones tied to the accessibility info in terms and conditions, which is your Art. 13 point.\nTracker with sources per country: https://www.complyloft.com/accessibility/eaa/enforcement\n\nSo there's a real deliverable in \"here's how we tested, here's what we covered and what we didn't\", sold as the evidence behind a vendor's ACR rather than as the ACR itself."}
