Skip to main content

Change log

A plain-language record of substantive changes to the Charter Mark certification scheme, the Human Values Framework, and our legal and governance pages. Each legal page's own version line is the authoritative record for that page — this is an index across all of them, newest first.

7 October 2026

  • Search: the explainer video page (/explainer) now tells search engines that its captions and transcript are available in English (UK and US), Irish (Gaeilge), French, German, Spanish, Italian, Dutch, Polish, Chinese and Japanese, in its search description, in an “Available in” line on the page (each language named in its own language), and in the video’s structured data, which now lists each caption file. Google has indexed the page and its video and has been asked to re-crawl it.
  • The explainer video page now has its own address in each caption language, so it can be found in searches in those languages: /explainer/en-us, /explainer/ga (Gaeilge), /explainer/fr, /explainer/de, /explainer/es, /explainer/it, /explainer/nl, /explainer/pl, /explainer/zh-hans and /explainer/ja, alongside /explainer in English (UK). Each version shows its page text, “About this video” and transcript in that language, starts the captions in that language, and links to all the others; search engines are told which version is in which language. The translations have not yet been reviewed by native speakers.
  • Pricing: the “Register” buttons on the pricing page now take visitors who aren’t signed in to the sign-up page (and back to registration afterwards) instead of showing “Page not found”. The organisation account page likewise sends signed-out visitors to sign in.
  • Security: the site’s Content Security Policy, which until now only reported problems, is now enforced. Browsers will only load scripts, frames and connections from the services the site actually uses (our sign-in provider, Stripe, our database and analytics), which limits the damage any injected script could do.
  • Accessibility: stronger text contrast on the How it works steps, the public register and the verify page; the “skip to content” link now works on the self-assessment page; and the site’s navigation areas are labelled for screen readers.
  • Search: shorter page titles and descriptions on the assessment, register, verify, pricing and consumer FAQ pages, so search results show them in full; the Advisory Board page now has its own title and a share image.
  • Explainer video page: the video’s preview image now loads first and the player no longer shifts the page while it loads. Mobile PageSpeed for /explainer rose from 80 to 93 (layout shift from 0.15 to 0); the homepage stays at 96 on mobile and 100 on desktop.
  • Each translated version of the explainer page now declares its own language to browsers (for example French on /explainer/fr), so screen readers, translation prompts and search engines that read the page treat it as that language.

6 October 2026

  • Google Search Console maintenance: two addresses for pages retired in August 2026 (/human-oversight and /responsible-ai, whose content now lives on the AI Transparency Statement) were reported to Google as “not found”. Both now redirect permanently to /ai-transparency, including from www.sacredlogic.org and our other domains. robots.txt now also tells search engines not to follow the link Cloudflare uses to protect e-mail addresses on our pages from spam bots; the protection itself stays on. Google has been asked to re-check the affected addresses. No changes to page content.
  • New page: the explainer video now has its own page at /explainer, so Google can list it in video search results (on the homepage it counts only as supporting content, so Google had not indexed it). The video plays in full 1080p HD on that page. The homepage video is unchanged.
  • Accessibility: the explainer video now has English captions (turn them on with the CC button in the video player), both on the homepage and on /explainer, and the /explainer page includes a full written transcript.
  • Caption languages: English (UK) captions now play by default every time the explainer video is opened, and a row of flags under the video (and at the top of the full-screen player on the homepage, where it fades out during playback) switches the captions to English (US), Irish (Gaeilge), French, German, Spanish, Italian, Dutch, Polish, Chinese (Simplified) or Japanese, or turns them off. Captions are now drawn by the site rather than the browser, so each one is placed in a clear space below the slide text (checked against every slide), stays clear of the player controls, and scales with the video. The translations were made from the English captions; “Charter Mark” and the “Research, Engage, Deliver” tagline stay in English.
  • On the /explainer page, the “About this video” text and the transcript now switch language together with the caption flags (and return to English when captions are turned off).
  • Mobile performance: the homepage now loads the caption files and caption controls only when the explainer video is opened. Including them on every homepage visit had delayed the main lighthouse image and lowered the mobile PageSpeed score from 98 to 87–89; it is back to 96–97. The site also no longer makes an unused early connection to a development database on every page.

3 October 2026

  • Site footer trade mark line shortened to the UK registration and the pending Irish and EU applications. The status of "Charter Mark" (an unregistered mark) continues to be stated in Terms of Service § 9 and the Disclaimer.

2 October 2026

  • Trade mark status updated: "Sacred Logic" is now a registered trade mark in the United Kingdom (UK00004411430, classes 9 and 42, entered on the register 2 October 2026). Terms of Service § 9, the Disclaimer, and the site footer now say so. Applications in Ireland and the EU remain pending, so the ™ symbol continues to be used in on-page copy; "Charter Mark" remains an unregistered mark.

25 September 2026

  • Mobile performance: the site stylesheet is now delivered inside each page instead of as a separate render-blocking file; the database client is loaded only on the pages that use it (assessment, register, verification, certificates, complaints and account pages) rather than on every page; the lighthouse glow is now a lightweight fade effect instead of a repainting filter animation; and the background and lighthouse glow begin pulsing a few seconds after the page has loaded, so the page settles faster on phones.
  • Search: the homepage explainer video’s structured data now gives its upload date as a full date and time with a timezone, fixing the “uploadDate is missing a timezone” notice in Google Search Console so the video can be shown with richer search features.
  • Automatic renewal: Charter Mark subscriptions run for 12 months and renew automatically each year, with the full annual fee charged in advance on the renewal date. Terms of Service §5 (updated 25 September 2026) now sets out the annual term, automatic renewal, a reminder before each renewal, at least 30 days’ notice of any price change, how to turn auto-renewal off (effective at the end of the year already paid for), and what happens if a renewal payment fails. The pricing page, FAQ, Stripe checkout page and renewal-reminder e-mail explain the same terms.
  • New on the organisation account page: a “Subscription & auto-renewal” panel showing whether your subscription will renew and when, with a “Manage billing & auto-renewal” button that opens Stripe’s secure billing page to turn auto-renewal off or on, update your card and download invoices.
  • Pricing: each tier now states that the annual fee is paid in full, annually in advance, at checkout — there are no monthly instalments.
  • Pricing checkout fix: returning to the pricing page with the browser Back button after starting a Stripe checkout no longer leaves the buttons stuck on “Redirecting...” — you can now switch currency or tier and check out again.
  • Proof-document template pack updated to v1.2: evidence is uploaded in your organisation account (Organisations → Account → Evidence documents), or e-mailed to [email protected].
  • Contact address: the AI Transparency Statement, Standards & References and Trademark Policy pages now direct legal and compliance questions to [email protected] (previously [email protected]).

24 September 2026

  • Pricing in three currencies: Charter Mark prices remain set in euro (our base currency), and fixed prices in pound sterling and US dollars are now also offered — Bronze €1,400 / £1,200 / $1,600, Silver €5,500 / £4,750 / $6,250, Gold €12,500 / £10,750 / $14,250 per year, with Platinum and the one-off certification issuance fee shown in each currency. The pricing page has a € / £ / $ switcher (defaulting to sterling for UK visitors and US dollars for US visitors) and customers are charged in the currency they select. These are fixed prices set by Sacred Logic, not live exchange-rate conversions, and are reviewed periodically.
  • Terms of Service (updated 24 September 2026): the certification issuance-fee clause now refers to the equivalent fixed sterling or US dollar amount shown on the pricing page, alongside the euro range.
  • Organisations FAQ: new question — “Which currencies can I pay in?”.
  • Pricing page improvements: tier prices and terms now load immediately with the page (previously they appeared only after the page finished loading in the browser), so they are visible to search engines and link previews.
  • Navigation: a Pricing link has been added to the footer of every page on the main site and the public register, and the Advisory Board page now carries the standard site footer.

22 September 2026

  • Google Search Console maintenance: submitted a refresh request via Google’s Outdated Content Removal tool for a dead domain (d15589-2.icart.cc) that had been incorrectly cited in Google’s AI Overview for ‘Sacred Logic’. Submitted AI Overview feedback to flag the broken link directly to Google. Requested priority re-indexing of sacredlogic.org via Search Console URL Inspection (‘Indexing requested’ confirmed). No site content changes. Remediation is Google-side only; propagation expected within days to weeks.
  • Performance optimisation (mobile LCP and initial bundle): stopped the Fraunces display-serif font from preloading — on mobile the font was racing the LCP lighthouse hero image for bandwidth; removing the preload hint lets the image win and load first. Removed the unused variable-font SOFT axis (controls letterform softness, not referenced in CSS) to reduce font file size. The floating chat widget (ChatWidget) is now deferred via next/dynamic with ssr:false, removing its JavaScript chunk from every page initial bundle. Root cause of the 22 Sep mobile score regression: the dark/light mode commit of 21 Sep added ~5 KB to globals.css and a synchronous FOUC-prevention script in head — these font and bundle fixes compensate on the bandwidth side.

21 September 2026 (a11y fix)

  • Accessibility fix (light mode): brand gold #b08d3e failed WCAG AA contrast on the light background (#f4f6fc) at only 2.89:1 — well below the 4.5:1 minimum for normal text. Introduced a new --color-brand-gold-text CSS token that equals #b08d3e in dark mode (no visual change) and switches to a darker accessible gold #7a6218 (≈5.0:1) in light mode. Applied to all gold text on the Framework, How It Works, and Fast Track pages: section eyebrows, step numbers, underline links, status badge text, and headings. Purely decorative icons remain unchanged as non-text elements require only 3:1 contrast.

21 September 2026 (dark/light mode)

  • Added dark/light mode toggle: a sun/moon icon button in the site header (PageHeader) switches between the existing dark theme and a new light palette. Theme preference is saved to localStorage (key: sl-theme) and restored on the next visit via a synchronous inline script in app/layout.tsx that runs before React hydrates, preventing any flash-of-wrong-theme. The light palette uses light blue-grey backgrounds (#f4f6fc) with white panels, dark text (#0a0f1e), and a blue focus ring — all defined as CSS custom property overrides under [data-theme="light"] in globals.css. No component files needed structural changes: every colour already flowed through CSS custom properties. A new --color-brand-navy-ink token (always dark navy, both modes) was added to keep button text legible on gold and white surfaces in the light theme.

21 September 2026 (rescore)

  • PageSpeed Insights rescore (live site, post GEO pass): mobile Performance 93/100, Accessibility 100/100, Best Practices 100/100, SEO 100/100, Agentic Browsing 3/3. Desktop Performance 100/100, Accessibility 100/100, Best Practices 100/100, SEO 100/100. Accessibility improvement from 96 to 100 confirmed — the <dl>→<div> fix for the FAQ accordion is reflected in this run. Desktop performance now showing 100/100.

21 September 2026

  • Added llms.txt at sacredlogic.org/llms.txt — a machine-readable index of the site's content, page purposes, key facts, and permissions, following the open llms.txt standard. Enables LLM-driven search engines (Perplexity, ChatGPT Search, Gemini, Claude) to efficiently parse and cite Sacred Logic content with correct context. Includes explicit guidance that certification status must be verified against the live public registry, not inferred from cached data.
  • Added TechArticle JSON-LD schema to /framework: cross-references the Organization and WebSite nodes already in the site-wide @graph; adds an about[] array of Thing nodes for AI safety certification, EU AI Act, human values in AI, and AI governance — improving citation accuracy in LLM search engines.
  • Added HowTo JSON-LD schema with five step nodes to /how-it-works: totalTime P14D, estimatedCost EUR 0, each step linked to its own page. Gives generative AI a structured representation of the five-step Charter Mark certification process.

21 September 2026 (earlier — scores)

  • PageSpeed Insights mobile score improved from 93 to 95/100; desktop 99/100. Confirmed via fresh PSI run after deploying the perf and accessibility fixes below.
  • Performance fix: removed mailto:[email protected] link and email address text from homepage HTML. Cloudflare's email-decode.min.js script was injected into every page that contained an email address, and on mobile it appeared in the critical render-blocking chain at approximately 970 ms. With no email addresses in the homepage HTML, Cloudflare no longer injects the script on this page. The CTA now links to /organisations.
  • Accessibility fix: changed the FAQ section wrapper element from <dl> to <div>. A definition list's children must be <dt>/<dd> pairs; wrapping <details>/<summary> accordion items in a <dl> is invalid HTML and was flagged by Lighthouse Accessibility. Corrects the markup without changing visible appearance or behaviour.

21 September 2026 (earlier)

  • SEO structured-data audit completed — overall search-engine signal score reached 99/100.
  • Added FAQPage JSON-LD schema to the homepage, making it eligible for Google's FAQ rich-result snippet in search. The schema covers five common questions about the Charter Mark, how certification works, EU AI Act alignment, registry verification, and cost.
  • Added an interactive FAQ accordion to the homepage ('Common Questions' section) using native HTML <details>/<summary> elements — no JavaScript required. Five questions matching the FAQPage schema, positioned after the 'Why AI Safety Certification Matters' prose section.
  • Added BreadcrumbList JSON-LD schema to all eleven public sub-pages, covering all main navigation routes and their nested FAQ pages. Pages covered: /about, /framework, /governance, /certification-rules, /organisations, /organisations/faq (two-level: Home > For Organisations > Certification FAQ), /consumers/faq (two-level: Home > For Consumers > Portal FAQ), /how-it-works, /fast-track, /consumers, /registry. Breadcrumb schema is now present on every significant page in the site's information architecture.
  • Added 'Why AI Safety Certification Matters' prose section to the homepage (~350 words) — explains the EU AI Act regulatory context, what independent AI safety certification means, the role of the public Sacred Logic Registry, and why the Charter Mark exists. Fills a content gap identified in the prior SEO audit.
  • Confirmed edge-caching performance: Cloudflare CF-Cache-Status HIT confirmed with TTFB 2–3 ms (down from 873 ms cold). Cache rule active: all incoming requests, Edge TTL 1 hour, ignore Cache-Control header from origin.
  • Confirmed all supporting technical signals: Twitter card set to summary_large_image with title and description; robots.txt references sitemap; sitemap.xml lists 30 public URLs.

14 September 2026

  • Mobile PageSpeed score recovered and improved to 94/100 (mobile) and 100/100 (desktop) after a dip to 88/100 caused by a brief regression.
  • Promoted the hero lighthouse image animation to its own compositor layer (will-change: filter) so the browser no longer repaints the page on every frame of the beacon-pulse glow effect. Eliminates the non-composited animation warning flagged by Lighthouse.
  • Enabled critical CSS inlining (Next.js experimental.optimizeCss with Critters): the browser now receives above-the-fold styles inline in the HTML rather than in a separate blocking stylesheet. Previously a 14.8 KiB CSS file was blocking page rendering for ~300 ms on mobile — it is now deferred and no longer in the critical path.
  • Added a .browserslistrc targeting modern browsers (last 2 years, > 0.5%, not IE 11), instructing the Next.js webpack bundler to skip polyfills for JavaScript features already supported natively. This removes a 16.4 KiB legacy polyfill chunk containing redundant implementations of Array.at, Array.flat, Object.hasOwn, and similar methods.

13 September 2026

  • Mobile PageSpeed score reached 96/100 (mobile) and 100/100 (desktop). Composited the HVF reading-progress bar animation — replaced CSS width transition with transform: scaleX() so the browser animates it on the GPU compositor thread without triggering layout, removing the non-composited animation warning. Fixed a build failure caused by experimental.browsersListForSwc, an option that was removed in Next.js 16.
  • Fixed a canonical URL mismatch that was preventing most pages from being indexed by Google. Since the 6 September release, the site's primary domain on Vercel had been changed from www.sacredlogic.org to sacredlogic.org (non-www), making sacredlogic.org the canonical address and www.sacredlogic.org a redirect to it. However, NEXT_PUBLIC_SITE_URL was still set to https://www.sacredlogic.org, so every page's canonical tag, sitemap URL, and social-share URL was pointing at the redirecting www version — causing Google to report 'Page with redirect' instead of indexing the page. NEXT_PUBLIC_SITE_URL has been corrected to https://sacredlogic.org. The sitemap submitted to Google Search Console has also been updated from the www URL to the canonical non-www URL. All 30 public pages are now correctly discoverable.
  • First-ever sitemap submission to Google Search Console. The sitemap at sacredlogic.org/sitemap.xml — which lists all 30 public pages with their priorities and update frequencies — had never been submitted. It is now submitted, processed successfully, and all 30 pages discovered.
  • Published a product user guide covering both audience journeys — organisations getting certified and consumers verifying AI. Documents all four Charter Mark tiers and fees, the 5-step certification process, the Compliance Fast Track, all ten Human Values Framework points in plain language, every public route, certificate verification, and the admin console.
  • Updated the site title, meta description, and social-share tags to reflect AI safety language — the fastest-growing search category in the AI space in 2026. The title now reads 'AI Safety Certification' rather than 'AI Trust Certification', and the meta description opens with the buyer's question — 'Verify AI is safe before you use it' — before explaining the certification scheme. This covers two audiences the previous wording missed: businesses searching for 'AI safety certification' and people searching for 'safe AI tools' or 'how to verify AI is safe'.
  • Added nine new keyword signals to the site: AI safety certification, AI safety standards, trustworthy AI, safe AI tools, verify AI safety, AI safe by design, AI accountability, AI transparency certification, and AI safety compliance. These reflect the terms most searched in connection with AI safety in 2026, identified by reviewing global search trend data.
  • Google PageSpeed SEO score: 100 out of 100.

7 September 2026

  • Improved homepage loading speed on mobile: the explainer video previously pre-loaded metadata (and partly the video file itself) even before anyone clicked to watch it, consuming significant bandwidth on slow mobile connections. It now waits until the play button is clicked before downloading anything — the video poster image included. On mobile networks this reduces the time before the main page content is fully visible.
  • Added a WebP version of the explainer video poster image (43% smaller than the JPEG original, at 1280×720 — plenty for a full-screen poster) so it loads faster when the user does click to watch.
  • Moved Clerk authentication to load only on pages that need it (sign-in/up, account pages, consumers, organisations, pricing, admin). Previously Clerk's JavaScript (~285 KiB) loaded on every page including the homepage; it now loads only when a visitor navigates to a page that shows sign-in buttons or requires an account.
  • Fixed the hero image loading speed on mobile: the lighthouse image was being served at full resolution (over 400 pixels wide) even on phones where it displays at roughly 160 pixels. It now receives an appropriately-sized image, which reduces the amount of data mobile visitors need to download before the page finishes rendering.
  • Homepage converted to a server-rendered page: the lighthouse hero image now has its preload instruction sent by the server in the initial HTTP response, so the browser starts downloading it immediately rather than waiting to discover it in JavaScript. This is the recommended approach for above-the-fold images and is one of the highest-impact changes for Largest Contentful Paint.
  • Fixed a layout jitter and performance bug on the Human Values Framework page: the scroll-progress bar at the top of the page was recalculating the document's full height on every scroll event — dozens of times per second — causing the browser to repeatedly interrupt rendering to recalculate layout. The dimensions are now cached and the bar updates inside a single animation frame, eliminating the jitter.
  • Removed a hidden preload hint for the Geist Mono monospace font. The font is used only in code-style elements below the fold and was never visible before the page had fully loaded, yet a preload instruction was asking the browser to fetch it as a high priority, blocking other resources. Removing it cuts one render-blocking request.
  • Added a preconnect hint for the database backend (Convex). The browser now opens the TCP and TLS connection to the data server in parallel with page rendering, rather than waiting until JavaScript requests data. This reduces the time before live registry data can appear.
  • Google PageSpeed (Lighthouse) performance score improved from 83 to 97 out of 100.

6 September 2026

  • Fixed two pages that had hard-coded non-www canonical URLs (advisory-board, complaints), causing Google to see a domain mismatch between the canonical it was given (sacredlogic.org) and the page it actually crawled (www.sacredlogic.org). Both now use relative canonical paths that resolve correctly against the site's canonical www domain.
  • Fixed a typo in the Advisory Board page's Open Graph metadata: the URL read sacredloic.org (missing the "g") instead of sacredlogic.org. This produced invalid structured data for that page.
  • Removed a redundant static robots.txt file from the public directory. The site correctly uses a dynamic Next.js App Router robots handler (app/robots.ts) that includes proper crawl-disallow rules for private routes; the static file was dead code that added ambiguity.
  • Corrected the site-wide canonical URL environment variable in the Vercel dashboard: NEXT_PUBLIC_SITE_URL was set to https://sacredlogic.org (non-www), while the site is served at https://www.sacredlogic.org. Because all relative canonical paths (e.g. /about, /pricing) are resolved against this variable, every page on the site was pointing search engines to the non-www version — which itself 301-redirects to www, creating a redirect loop in Google's eyes and causing the "Alternate page with proper canonical tag" indexing error across the board. The variable is now correctly set to https://www.sacredlogic.org.

5 September 2026

  • The "What is SACRED LOGIC™?" homepage explainer has moved from a YouTube embed to a self-hosted video, so watching it no longer depends on a third-party video platform. The video itself was also refreshed with updated end-user-facing copy about searching the free public registry.
  • Fixed a page that had accidentally been hidden from search engines: /pricing was made publicly visible without sign-in on 22 August 2026 (see that entry below), but a leftover technical setting kept telling search engines not to index it. That has now been corrected, and the page has its own proper title, description, and preview card.
  • Fixed the sitemap (the technical index that helps search engines discover every page on the site): nine real, public pages were missing from it entirely, including the pricing page, the "How it Works" page, and the Compliance Fast Track page. All are now included.
  • Shortened around a dozen page descriptions that were too long to display fully in search results, including on the Trademark Policy, Data Processing Agreement, Disclaimer, AI Transparency, Certification Rules, and Cookie Policy pages, while keeping the important legal meaning of each.

4 September 2026

  • All pre-launch and not-yet-operational language has been removed from public-facing pages. Affected pages: homepage (ticker and Fast Track CTA), Security & Vulnerability Disclosure, Consumer Registry, Consumer Consent Notice, Governance & Independence, Accessibility, and the Data Processing Agreement.
  • Security page: removed the "Honest note: Sacred Logic™ is a pre-launch MVP" paragraph and the "pre-launch platform" qualifier from the security summary. The Independent Assessment section is now written as a commitment; the Secure Reporting section no longer uses "don't yet" framing.
  • Consumer Registry: removed the notice stating that additional compliance data sources "are coming soon". Removed the "(coming soon)" label from the disabled Request an Introduction button.
  • Consumer Consent Notice: removed the placeholder block stating the portal and supplier partnerships were "not yet operational".
  • Governance & Independence: replaced all "commercial pilot" references with present-tense operational language. The governance roadmap now describes a scheme that is growing and maturing rather than preparing to launch.
  • Accessibility: replaced "This site has not yet undergone a formal accessibility audit" with a positive statement of existing groundwork and commitment to a future audit.
  • Data Processing Agreement: removed "(in formation)" from the Sacred Logic Ltd. legal entity name.
  • Homepage news ticker: removed the "Certification Launch Partner's innovation programme now open" and "Coming soon: integrated live EU compliance data" ticker items. Updated the Fast Track sub-caption from "Launch Partner programme open — early access, reduced rate" to "Certification open — contact [email protected] to get started".

28 August 2026

  • The site title and social-share tags have been corrected: "SACRED LOGIC™" has been changed to "Sacred Logic" across the <title>, og:title, and twitter:title fields. The trademark symbol (™) was being converted to the text "TM" by DuckDuckGo's snippet engine, producing "SACREDLOGICTM" in search results. The symbol is preserved in all on-page copy and headings — removed from metadata tags only.
  • A full mobile-navigation hamburger menu has been added to the site header. On small screens the desktop nav is replaced by a 44×44px hamburger button that opens a full-width animated drawer. Tapping any link in the drawer closes it automatically. The desktop navigation is unchanged.
  • Footer navigation links now have larger touch targets (minimum 44px) and extra bottom padding to prevent the live-chat widget from overlapping navigation links on mobile.
  • Homepage navigation is now visible on the homepage (previously hidden) and the search/enquiry form stacks vertically on small screens.
  • The homepage trust-row wording has been updated: "Human-reviewed by a named, independent assessor" is now "Human-reviewed by a named assessor operating under published independence safeguards", and the accompanying governance link now reads "See how certification independence works →". This more precisely describes the current sole-founder structure without implying a third-party body that does not yet exist.
  • Certification Rules and Terms of Service: "Sacred Logic™ is an independent certification body" replaced with "Sacred Logic™ operates an independent, founder-led certification scheme" in both documents. The new wording preserves the important meaning — Sacred Logic is independent of the organisations it certifies — while accurately describing the current governance model.
  • Fast Track page: two explicit section headers added to the Human Values Framework points list, clearly separating Points 1–7 ("accepted on the basis of declared EU AI Act conformity and documentary evidence; not independently re-audited by Sacred Logic™") from Points 8–10 ("assessed by Sacred Logic™ against the Human Values Framework"). This eliminates any ambiguity an auditor or sophisticated buyer could identify between the Fast Track route and the canonical framework.
  • Machine-readable wordmark fix (complete): the "Sac red Logic™" split — where adjacent colour spans caused crawlers and accessibility trees to read two separate tokens — has now been resolved across all five instances site-wide: homepage H1, homepage disclaimer, certificate page, About page, and the assessment download report. Each instance now renders "Sacred" as a single text node using a CSS background-clip gradient, preserving the visual green-to-red split while exposing one clean word to every automated reader.
  • Article 95 wording standardised: two pages that described the scheme as a "voluntary AI assurance self-assessment consistent with Article 95" have been updated — Terms of Service now reads "voluntary AI assurance questionnaire consistent with Article 95" (accurate for the tool) and the Organisations landing page now reads "voluntary AI assurance scheme consistent with Article 95", matching the canonical homepage formulation. The Charter Mark requires human validation and is not a self-assessment.

22 August 2026

  • The Charter Mark tiers and pricing page (/pricing) is now publicly visible without requiring registration — visitors can see Bronze (€1,400/yr), Silver (€5,500/yr), and Gold (€12,500/yr) tier prices and a Platinum band (€25,000–€80,000/yr) before committing to creating an account. The interactive checkout flow is preserved and shown once a visitor is logged in with a registered organisation account.
  • A governance trust signal has been added to the homepage: a line directly below the main trust row reads 'Human-reviewed by a named, independent assessor — read our governance statement →', linking to /governance. This gives enterprise buyers a one-click path to assessor credentials and independence safeguards without leaving the hero section.

19 August 2026

  • The 'Close ✕' label on the About page header has been changed to '← Home', removing ambiguity for visitors who arrive directly from search — the link has always gone back to the home page, and the label now says so plainly.
  • The site's canonical URL is consistently www.sacredlogic.org, managed at the infrastructure level via Vercel's domain settings — ensuring a single canonical URL across the site and avoiding any potential for duplicate-content indexing.
  • Fixed a long-standing visual defect where the Sacred Logic™ wordmark rendered as 'Sac red Logic™' with a spurious space in 'Sacred' on the homepage, certificate page, registry shell, and Charter Mark badge — caused by JSX whitespace between adjacent colour-split spans.
  • The homepage footer disclaimer now reads 'The Sacred Logic™ Charter Mark is a voluntary Article 95 certification' rather than 'self-assessment', accurately reflecting that the Charter Mark requires human validation by a Sacred Logic™ assessor.
  • The homepage news ticker has been simplified: the Compliance Fast Track announcement has been removed (it is already shown as a prominent pill on the same page) and the two near-identical EU AI Act implementation-date items have been merged into one.
  • The public register now shows a smooth loading animation while entries are being fetched, and displays a clear message with a 'Clear all filters' action when a search returns no results.
  • Fast Track evidence basis is now colour-coded on both the register list and each organisation's register profile: CE mark (green), ISO/IEC 42001 (blue), and self-declared (amber) — so a reader can see at a glance how strong the underlying compliance evidence is without reading the prose.
  • The How it Works page (/how-it-works) now displays a distinctive icon and estimated time for each of the five certification steps, making the journey more tangible before you start.
  • The certificate verification page (/verify) now clarifies that 'no certificate found' means the ID was not found in any status — including expired, suspended, and revoked — so a reader knows the search was exhaustive.
  • Fixed a bug where the self-assessment Part A risk questions shared a single radio group, causing selecting one answer to deselect another.
  • The 'Compliance Fast Track' label on registry entries now has a tooltip keyboard and screen-reader users can reach, explaining what the fast track covers and what it does not.
  • Fixed a bug where the 'public register entry' text in the Fast Track disclosure on the electronic certificate was a non-clickable label rather than a link — it now navigates to the organisation's registry profile.
  • The keyboard-accessible 'Skip to main content' link now works on every page of the site, not only the homepage — the target anchor was missing from the main element on all other pages.
  • A dedicated Compliance Fast Track page (/fast-track) is now published, explaining who qualifies, what Sacred Logic™ assesses (Human Values Framework points 8–10 only), what is not covered (points 1–7), the evidence basis options (CE mark, ISO/IEC 42001, or self-declared), the step-by-step process, and the required points-1–7 disclosure. The homepage Fast Track pill now links directly to this page.
  • Removed two internal drafting notes from the Certification Rules page: the '(to be finalised before the pilot)' parenthetical from the complaints acknowledgement timescales, and the end-of-page notice describing the document as a good-faith operational description. Both are now clean, unqualified policy text.
  • Fixed a text-extraction and accessibility issue on the homepage where 'Article 95' and 'certification' appeared joined without a space in the Article 95 disclaimer.
  • The Part B evidence cards on the self-assessment now use progressive disclosure: the detailed evidence-requirement text is hidden behind a 'What's covered?' toggle so the initial view is less daunting, while the full detail remains available on demand.
  • Published a new Governance and Independence page (/governance) explaining who can assess an application, who makes certification decisions, how conflicts of interest are handled, the structural safeguards protecting decision integrity, appeals, and the scheme's commitments and their limits. Linked from the site footer.
  • Following completion of the external legal review of the whole legal suite, the 'not yet reviewed by a lawyer' notices previously shown at the foot of the Cookies, Disclaimer, and Security pages have been removed.

10 August 2026

  • After completing the self-assessment and opting in, visitors now receive a personalised copy of their results by email — including their readiness score, recommended Charter Mark tier, a Human Values Framework point-by-point breakdown, EU AI Act risk profile, and next-step guidance.
  • The results email now includes the Sacred Logic lighthouse image and branded tagline, visible even when email clients block external images.
  • The follow-up form now accepts an optional name and phone number, which appear in both the visitor copy and the certification team notification.
  • A one-click link to request a complimentary 15-minute Microsoft Teams video briefing with the certification team is included in every visitor results email.
  • Fixed a bug where a high readiness score could incorrectly display as 'Not ready for certification yet' in the visitor copy email.

4 August 2026

  • Fixed a defect where opting in for follow-up after completing the self-assessment didn't reliably notify the certification team.
  • Footer navigation links now have a larger touch/click target, meeting the WCAG 2.5.8 minimum target size.

1 August 2026

  • Certification Rules gained a named lifecycle section (issue / suspension / investigation / appeal / reinstatement / expiry / withdrawal) and a reinstatement paragraph distinct from revocation.
  • The certificate and public verification pages now show the framework version a mark was assessed under.
  • Misconduct reports can now genuinely be submitted anonymously — name and e-mail are optional end-to-end, not just described as optional.
  • The personal-data retention schedule is now also published as a plain reader-facing table, alongside a new “deleted accounts” entry.
  • Revoking a Charter Mark no longer deletes the record outright: /verify and the public register profile now distinguish Current, Suspended, Revoked, and Expired, so a revoked mark is no longer indistinguishable from one that never existed.
  • Every certification decision (issue, certify, certify with conditions, defer, decline, revoke) now records the framework version applied and the evidence on file in the internal audit trail, and a reviewer can attach a rationale note to any decision, not only a refusal.
  • The Trademark and Brand Usage Policy now names the Charter Mark licence explicitly (non-exclusive, revocable, non-transferable), and adds terms on comparative use, advertising and social media, and what happens on expiry, alongside suspension and revocation.
  • The public certificate-status badge's cache lifetime was shortened (1 hour to 5 minutes) so a suspended or revoked mark stops showing a cached “Certified” badge sooner.
  • The Security page now discloses plainly that there is no PGP key or public status page yet, rather than being silent on the point.
  • Every legal page now shows a “last reviewed” date alongside its version and last-updated date.
  • Fast-track Points 1–7 third-party-reliance disclosures were added to every registry-facing page (registry list, registry profile, /verify, and the printable certificate) — previously only the registry profile page had one — and the certificate's headline claim is now conditional on the actual assessment route.
  • /admin/chat-logs access was audited end to end (server-verified admin identity, edge gating, salted non-reversible visitor hashing) — no exposure found.
  • A Data Processing Agreement is now published at /dpa and applies automatically from registration, rather than existing only as an unexecuted internal draft.
  • Every governing-law reference site-wide now names the Republic of Ireland specifically, and the Terms of Service and Data Processing Agreement carry an explicit, irrevocable exclusive-jurisdiction clause.
  • The Terms of Service gained an explicit strict-liability disclaimer for a Compliance Fast Track organisation's self-declared Points 1–7 conformity.
  • Fixed a consumer-portal consent-flow defect: declining the optional marketing-lead-sharing consent had incorrectly blocked use of the free portal. The two consents (data processing, required; marketing-lead sharing, optional) are now genuinely independent, matching the feature's own Data Protection Impact Assessment.
  • The self-assessment results screen now states explicitly that the score is calculated by fixed rule-based logic, not by an AI model.
  • The Disclaimer page's certification-scope section now carries the same Points 1–7 fast-track disclosure as the registry pages.

31 July 2026

  • Published this AI Transparency Statement page and split the complaints process into three tracks: operational complaints, certification appeals, and misconduct reports (Certification Rules § 8).
  • Added a “who reviews applications today” disclosure to Certification Rules § 4, and a version number alongside the effective/last-updated date on every legal page.
  • Clarified that Sacred Logic acts as a data controller for most of the site but as a processor for an organisation customer's certification application data specifically, on that customer's instructions (Privacy Policy, Data Processing Agreement template).
  • Added a dedicated Electronic Marketing (PECR / ePrivacy) section to the Privacy Policy, covering e-mail, phone, and SMS marketing consent under Ireland's ePrivacy Regulations.
  • Added explicit “not legal advice / not a regulatory approval / not a guarantee of ongoing conduct” disclosures to the certificate, registry, and verification pages, and finalised the personal-data retention schedule.

25 July 2026

  • Published a standalone Cookie Policy and Certification Rules page (methodology, decision rules, validity/renewal, suspension & withdrawal, complaints & appeals).
  • Substantially expanded the Privacy Policy (international transfers, complaint routes to the Irish DPC / UK ICO, an automated-decision-making statement, a security section) and the Terms of Service, following a first external legal review of the whole legal suite.
  • Drafted a Data Protection Impact Assessment for consumer lead-sharing and a customer Data Processing Agreement template.

24 July 2026

  • Published a Security page with a coordinated vulnerability-disclosure process, and a standalone Disclaimer page.
  • Strengthened limitation-of-liability and warranty language, including an explicit liability cap.
  • Site-wide accessibility pass: colour-contrast fixes across the certificate, registry, and admin pages.

21 July 2026

  • Expanded the Terms of Service and rebuilt the public register as a standalone, plain-language public page.
  • Introduced the electronic (printable/PDF) Charter Mark certificate.

18–20 July 2026

  • Published the initial Terms of Service and Privacy Policy, and the consumer consent notice for the B2C portal.

This log covers substantive, public-facing changes only — not every internal edit. Issued Charter Marks keep the framework version they were assessed under (see Certification Rules § 10).