About & Privacy

About this service

ScamPrism is operated by Paul Laikind as a personal, non-commercial project under the PKL Media Lab brand. It is provided free of charge, on an “as is” basis, without warranties of any kind.

Access is invitation-only. ScamPrism is not a public product and is not promoted as one.

By using this service, you acknowledge that

• Output is generated by artificial intelligence and may be inaccurate, incomplete, or biased.
• The service is informational only and is not a substitute for professional advice (legal, financial, medical, or otherwise).
• A “No Clear Indicators” result is not a guarantee of safety. ScamPrism never asserts that a message is safe. Always treat any message that asks for credentials, money, or personal information with skepticism, regardless of the result.
• You are responsible for verifying any information before acting on it.
• The operator is not liable for any decisions made or actions taken based on the service’s output.
• The service may be modified, paused, or discontinued at any time without notice.

What is stored

Each analysis produces a report containing the submitted message text, the extracted scam signals, the threat score, and the verdict. Reports are stored temporarily on the application server and accessible via a unique, unguessable URL for 48 hours, after which they are automatically and permanently deleted.

For email submissions, the email address of the person who forwarded the message is stored solely to deliver the reply notification. It is deleted along with the report after 48 hours.

What is not stored

No user accounts are created beyond the access codes the operator issues. No tracking cookies are set. No analytics about message content are aggregated across reports. Reports are not indexed for search and are not made publicly browsable. Data is not sold, shared with third parties beyond the processors named below, or used for purposes beyond operating the service.

Third parties involved in processing

• Anthropic — message text is sent to the Claude API for signal extraction and verdict generation. Per Anthropic’s API terms, API content is not used for model training by default.
• Google Safe Browsing — URLs extracted from analyzed messages are checked against Google’s Safe Browsing threat database to identify known phishing and malware links. Only URLs are sent, not message content.
• Brave Search — short, derived search queries (not full message text) are sent to Brave Search for signal verification.
• CloudMailin — receives forwarded emails and posts the parsed content to ScamPrism’s webhook.
• Resend — sends reply emails for the forwarded-email path.
• ImprovMX — relays inbound email from forward@scamprism.com and register@scamprism.com to CloudMailin.
• Cloudflare — DNS and TLS termination for scamprism.com.
• Railway — application hosting.

Operator access to your data

The operator (Paul Laikind) does not have access to the content of messages you submit for analysis, the email addresses of senders in forwarded emails, or which access code was used for any specific analysis. The operator administration panel displays only aggregate operational data: timestamp, source type (email or paste), threat score, verdict, and cost per analysis. No message content, sender addresses, or user-identifying information is accessible through the administration interface.

Analysis reports — including the submitted message text, extracted signals, score, and verdict — are automatically deleted after 48 hours. This deletion is permanent and not reversible. No message content is retained beyond that period.

Summary metadata (score, verdict, timestamp, analysis cost) may be retained in aggregate form for service monitoring. This metadata contains no message content and cannot be used to reconstruct the original message.

To request early deletion of a specific report, contact the operator with the report URL at contact@scamprism.com.

Contact

Questions about this service or this policy can be sent to contact@scamprism.com.