MarkusNeusinger
diff --git a/‎api/routers/seo.py‎
Lines changed: 4 additions & 2 deletions b/‎api/routers/seo.py‎
Lines changed: 4 additions & 2 deletions
diff --git a/‎app/src/pages/LegalPage.test.tsx‎
Lines changed: 43 additions & 0 deletions b/‎app/src/pages/LegalPage.test.tsx‎
Lines changed: 43 additions & 0 deletions
diff --git a/‎app/src/pages/LegalPage.tsx‎
Lines changed: 43 additions & 21 deletions b/‎app/src/pages/LegalPage.tsx‎
Lines changed: 43 additions & 21 deletions
diff --git a/‎changelog.d/legal-privacy-audit.md‎
Lines changed: 16 additions & 0 deletions b/‎changelog.d/legal-privacy-audit.md‎
Lines changed: 16 additions & 0 deletions
diff --git a/‎tests/unit/api/test_routers.py‎
Lines changed: 8 additions & 0 deletions b/‎tests/unit/api/test_routers.py‎
Lines changed: 8 additions & 0 deletions
@@ -1054,8 +1054,10 @@ async def seo_libraries():
     '<a href="https://x.com/MarkusNeusinger">X</a>). Full legal notice with '
     'contact details on the <a href="https://anyplot.ai/legal">interactive page</a>.</p>'
     "<h2>Privacy</h2>"
-    "<p>Analytics: Plausible Analytics (EU, proxied) — no cookies, no personal data collected. "
-    "Hosting: Google Cloud Run (Netherlands).</p>"
+    "<p>Analytics: Plausible Analytics (EU), served over a path on this domain — no cookies, no "
+    "cross-site identifier, no IP address stored. Server logs are kept 30 days; a feedback entry "
+    "is kept until it is deleted by hand and is readable only by the operator. Hosting: Google "
+    "Cloud Run (Netherlands), with Cloudflare at the edge.</p>"
     "<h2>Transparency</h2>"
     "<p>The whole stack — specs, pipeline, API and frontend — is open source at "
     '<a href="https://github.com/MarkusNeusinger/anyplot">github.com/MarkusNeusinger/anyplot</a>; '
 
@@ -50,6 +50,49 @@ describe('LegalPage', () => {
     expect(plausibleLinks.length).toBeGreaterThan(0);
   });
 
+  // The privacy section's load-bearing claims, each pinned where it is true.
+  // A shortening pass is exactly what drops a qualifier — "30 days" silently
+  // spreading over a store that has no timer, or the objection right losing
+  // the condition that makes it one.
+  it('names the legal basis, the jurisdiction and the whole set of rights', () => {
+    render(<LegalPage />);
+
+    expect(screen.getByText(/legitimate interest in protecting the site/)).toBeInTheDocument();
+    expect(screen.getByText(/swiss data protection law applies/)).toBeInTheDocument();
+    expect(
+      screen.getByText(/object to the processing on grounds relating to your particular situation/)
+    ).toBeInTheDocument();
+    expect(screen.getByText(/complain to a supervisory authority/)).toBeInTheDocument();
+  });
+
+  it('gives each store its own retention, and no store a borrowed one', () => {
+    render(<LegalPage />);
+
+    // Cloud Logging has the 30 days; the feedback table has no timer at all.
+    expect(screen.getByText(/retained for 30 days/)).toBeInTheDocument();
+    expect(screen.getByText(/nothing deletes them on a timer/)).toBeInTheDocument();
+  });
+
+  it('does not claim more privacy than the code delivers', () => {
+    render(<LegalPage />);
+
+    // The feedback widget asks for "Name or email (optional)", so the old
+    // blanket "no personal data" was false — and so is any restatement of it:
+    // an IP address, an IP hash and the feedback session id are personal data
+    // whether or not you typed them, so the bullet promises only name/email.
+    expect(screen.getByText(/no name or email unless you type one/)).toBeInTheDocument();
+    // Everything the feedback record actually holds, so a later trim cannot
+    // quietly drop the two identifiers a reader would care about most.
+    expect(screen.getByText(/your window size/)).toBeInTheDocument();
+    expect(
+      screen.getByText(/a random id that ties your own submissions together/)
+    ).toBeInTheDocument();
+    // Plausible's real property is cookieless and identifier-free, not that
+    // the script is ours — it is Plausible's, only served from our domain.
+    expect(screen.getByText(/the script is theirs/)).toBeInTheDocument();
+    expect(screen.getByText(/Cloudflare stands in front of the site/)).toBeInTheDocument();
+  });
+
   it('renders the technology stack', () => {
     render(<LegalPage />);
 
 
@@ -188,22 +188,26 @@ export function LegalPage() {
 
           <Box sx={{ maxWidth: 760, mx: 'auto' }}>
             <Typography sx={subheadingStyle}>data controller</Typography>
-            <Typography sx={textStyle}>Markus Neusinger (see legal notice above)</Typography>
+            <Typography sx={textStyle}>
+              Markus Neusinger (see legal notice above). the ground for all of it is an
+              operator&apos;s legitimate interest in protecting the site and knowing how often it is
+              read; swiss data protection law applies, and the GDPR on top for visitors from the EU.
+            </Typography>
 
             <Typography sx={subheadingStyle}>what we collect</Typography>
             <Typography sx={textStyle}>
-              <strong>anonymized analytics.</strong> we use{' '}
+              <strong>anonymized analytics.</strong> the counting is done by{' '}
               <Link href="https://plausible.io" target="_blank" rel="noopener" sx={proseLinkStyle}>
                 Plausible Analytics
               </Link>
-              , a privacy-focused analytics tool. it collects no personal data, uses no cookies, and
-              does not track you across websites. we track: page views, navigation patterns, code
-              copies, image downloads, search queries, filter usage, UI interactions (tab toggles,
-              theme preference, banner dismissals), and anonymized performance metrics (Core Web
-              Vitals: LCP, CLS, INP) to keep the site fast. when you share a link, we read the
-              requesting bot&apos;s user-agent to detect the platform (e.g., LinkedIn, WhatsApp) —
-              no data about the eventual viewer is collected at that step. all data is aggregated
-              and anonymous.
+              . the script is theirs, but it is served over a path on this domain: no cookies, no
+              identifier that would recognize you on other sites, no IP address kept there, and your
+              browser talks only to us. counted are page views, search queries and filter usage,
+              code copies and image downloads, theme and navigation clicks, and Core Web Vitals
+              (LCP, CLS, INP) to keep the site fast. when you share a link, we read the requesting
+              bot&apos;s user-agent to detect the platform (e.g., LinkedIn, WhatsApp) — no data
+              about the eventual viewer is collected at that step. all data is aggregated and
+              anonymous.
             </Typography>
             <Typography sx={textStyle}>
               <strong>public dashboard.</strong> our analytics are{' '}
@@ -230,16 +234,31 @@ export function LegalPage() {
               </Link>{' '}
               for security and debugging purposes.
             </Typography>
+            <Typography sx={textStyle}>
+              <strong>feedback.</strong> the feedback widget keeps what you send it: the message,
+              the reaction, anything you put in the optional contact field, the page you were on,
+              your window size, your user-agent, a random id that ties your own submissions
+              together, and a one-way hash of your IP address that limits spam. only the operator
+              can read those entries, and nothing deletes them on a timer — ask and yours is gone.
+            </Typography>
+            <Typography sx={textStyle}>
+              <strong>the edge.</strong> Cloudflare stands in front of the site and processes your
+              IP address there to rate-limit, to filter bots — which can put a short check page in
+              front of a request — and to absorb DDoS attacks.
+            </Typography>
 
             <Typography sx={subheadingStyle}>what we do not collect</Typography>
             <Typography sx={textStyle}>
               • no user accounts or personal profiles
               <br />
-              • no personal data (names, emails, etc.)
+              • no name or email unless you type one — the optional contact field in the feedback
+              widget is the only place that asks
               <br />
-              • no cookies at all (we use localStorage for UI preferences only)
-              <br />• <strong>no ai training</strong>: your interactions are not used to train ai
-              models
+              • no cookies at all (localStorage holds UI preferences — theme, image size — a cached
+              release number, and, once you use the feedback widget, the random id above)
+              <br />• <strong>no ai training</strong>: nothing you send is used to train ai models.
+              the ai in this project writes the catalogue, at build time; it never sees a
+              visitor&apos;s request
             </Typography>
 
             <Typography sx={textStyle}>
@@ -250,7 +269,9 @@ export function LegalPage() {
 
             <Typography sx={subheadingStyle}>hosting &amp; third parties</Typography>
             <Typography sx={textStyle}>
-              all services are hosted in the EU (Netherlands, europe-west4):
+              the site, its database and its images run in the EU (Netherlands, europe-west4);
+              Cloudflare&apos;s network is global, and Plausible is a service of its own on its own
+              EU servers:
             </Typography>
             <Table sx={{ ...tableStyle, ...firstColStyle }}>
               <TableBody>
@@ -279,13 +300,14 @@ export function LegalPage() {
 
             <Typography sx={subheadingStyle}>your rights</Typography>
             <Typography sx={textStyle}>
-              you have the right to access, rectify, erase, and export your data. since we do not
-              store personal data, there is typically nothing to delete or export. for questions,
-              contact{' '}
+              you can ask what is stored about you and have it corrected, erased or its use
+              restricted, object to the processing on grounds relating to your particular situation,
+              and complain to a supervisory authority. in practice that means a feedback entry or a
+              log line; a few lines to{' '}
               <Link href="mailto:admin@anyplot.ai" sx={proseLinkStyle}>
                 admin@anyplot.ai
-              </Link>
-              .
+              </Link>{' '}
+              are enough.
             </Typography>
           </Box>
         </Box>
@@ -554,7 +576,7 @@ export function LegalPage() {
             mt: 2,
           }}
         >
-          last updated: July 2026
+          last updated: September 2026
         </Typography>
       </Box>
     </>
 
@@ -0,0 +1,16 @@
+### Changed
+
+- **The privacy policy says what the code actually does, per store.** The section had drifted
+  behind a year of features: it promised "no personal data (names, emails, etc.)" while the
+  feedback widget asks for exactly that ("Name or email (optional)") and keeps it, with the
+  message, the page, the user agent and an IP hash, in a table nothing prunes on a timer. It
+  also had no legal basis, no jurisdiction, and a rights list missing restriction, objection
+  and the complaint to a supervisory authority. Each store now carries its own retention —
+  Cloud Logging's 30 days, a feedback entry until it is deleted by hand — because a single
+  blanket figure is the qualifier a shortening pass loses first. Plausible is described by the
+  property that matters (no cookies, no cross-site identifier, no IP stored, the browser
+  talking only to our domain) rather than the false one it used to claim (the script is
+  Plausible's, only served from here), Cloudflare's edge processing of the IP is named, and
+  the "all services in the EU" line no longer swallows a global CDN and a separate analytics
+  service. The crawler body in `api/routers/seo.py` mirrored the same wrong sentence and was
+  corrected with it (#11217).
@@ -1167,6 +1167,14 @@ def test_seo_legal_names_operator_and_privacy_facts(self, client: TestClient) ->
         assert "Plausible Analytics" in response.text
         assert "no cookies" in response.text
         assert "Google Cloud Run" in response.text
+        # The crawler body used to promise "no personal data collected", which
+        # the feedback widget's optional contact field made false. It now
+        # carries the same per-store retention the human page does.
+        assert "no personal data collected" not in response.text
+        # BOTH retentions, not only the logs' — pinning just the 30 days is how
+        # that figure silently spreads over the store it is not true for.
+        assert "kept 30 days" in response.text
+        assert "until it is deleted by hand" in response.text
 
     def test_seo_mcp_tells_agents_how_to_connect(self, client: TestClient) -> None:
         """The /mcp page's audience is AI agents — the bot body must carry the