Jul 20 - Update codes to comply with some framework (SOC 2 TYPE 2, ISO, etc)

app/routes/support.py

@@ -123,6 +123,33 @@ FAQS = [
 _KB_MAX_CHARS = 6000
 
 
+# ── PII redaction for the outbound Groq payload ───────────────────────────────
+# Groq is a third-party processor. Customers may type identifying details
+# (their own email/phone, or a coworker's) into a support question; there is
+# no need for that to leave the app to get a helpful, generic answer. This
+# scrubs a best-effort set of PII patterns from the copy of the text sent to
+# Groq only — the original text is still saved as-is in support_chat_messages
+# so the customer's own conversation history reads normally in the app.
+import re as _re
+
+_PII_PATTERNS = [
+    (_re.compile(r'[\w.+-]+@[\w-]+\.[\w.-]+'), '[redacted-email]'),
+    (_re.compile(r'\b\d{3}-\d{2}-\d{4}\b'), '[redacted-ssn]'),
+    (_re.compile(r'\b(?:\d[ -]?){13,19}\b'), '[redacted-number]'),
+    (_re.compile(r'\b(?:\+?1[ .-]?)?\(?\d{3}\)?[ .-]?\d{3}[ .-]?\d{4}\b'), '[redacted-phone]'),
+]
+
+
+def _redact_pii(text):
+    """Best-effort scrub of email/phone/SSN/card-like sequences from outbound text."""
+    if not text:
+        return text
+    redacted = text
+    for pattern, placeholder in _PII_PATTERNS:
+        redacted = pattern.sub(placeholder, redacted)
+    return redacted
+
+
 def _system_prompt_with_kb():
     """Return the base system prompt plus all ACTIVE admin knowledge entries
     (phase38), so staff can curate the chatbot's knowledge without code changes.
@@ -214,11 +241,13 @@ def chat_message():
             client = Groq(api_key=api_key)
 
             messages = [{'role': 'system', 'content': _system_prompt_with_kb()}]
-            # Append prior conversation (cap at last 20 turns to control token usage)
+            # Append prior conversation (cap at last 20 turns to control token usage).
+            # Redact PII-shaped text before it leaves the app for the Groq API —
+            # the unredacted originals stay in support_chat_messages below.
             for m in history[-20:]:
                 if m.get('role') in ('user', 'assistant') and m.get('content'):
-                    messages.append({'role': m['role'], 'content': m['content']})
-            messages.append({'role': 'user', 'content': user_message})
+                    messages.append({'role': m['role'], 'content': _redact_pii(m['content'])})
+            messages.append({'role': 'user', 'content': _redact_pii(user_message)})
 
             model = os.environ.get('GROQ_MODEL', 'llama-3.3-70b-versatile')
             completion = client.chat.completions.create(