AI-Assisted Content Detector — Proof of Concept

Technical prototype portfolio — demonstration environment

Use the bundled synthetic example below to inspect processing steps, review findings and the review flow. Do not submit personal, confidential or client-owned information. Contact Chrisian to discuss a controlled evaluation. A fixed set of automated detectors (SSNs, emails, phone numbers, Luhn-validated card numbers, profanity) always runs first; Claude then screens the content against the two custom criteria shown below, always flagged for a human to confirm or dismiss.

Synthetic sample content

sample-case-note.txt — the exact text run below.

Case intake note — initial client call summary.

Client reported ongoing dispute with former employer. Contact details on file:
jane.doe@example.com, reachable at (555) 123-4567. For identity verification
the client provided SSN 123-45-6789, confirmed against the retainer agreement
on file.

Client mentioned the case involves her son, Michael Doe, who is 14 years old
and was present during the incident described. Client was frustrated during
the call and said the employer's conduct was "absolute bullshit."

Settlement discussion: opposing counsel has informally proposed a settlement
of $85,000 to close the matter without further litigation, pending client's
review. A deposit of the retainer fee was processed using card number
4111 1111 1111 1111 as a placeholder for billing setup — to be replaced with
the client's actual payment method before filing.

Review criteria for this sample

Claude quotes the exact passage that matched, and every quote is checked against the real content before it's shown.

Names a Minor — Mentions a specific named individual who is under 18 years old

Discloses Settlement Amount — States a specific monetary figure discussed or proposed as a settlement