What Summarly does and doesn't do with your messages

What Summarly does

• Accepts pasted text, uploaded images and QR code photos
• Sends submitted content to an AI service (Anthropic's Claude API) to generate a plain-English explanation
• Returns a risk label, explanation of what the message appears to say, warning signs found, and suggested next steps
• Logs anonymous technical metadata (not the content of your submission) for quality and performance monitoring
• Provides free access during the public beta, with no account required

What Summarly does not do

• Does not store the content of messages you submit
• Does not confirm whether something is definitely genuine or definitely a scam
• Does not contact banks, organisations or authorities on your behalf
• Does not provide legal, financial, medical, tax or professional advice
• Does not replace official fraud reporting or emergency services
• Does not use advertising cookies or behavioural tracking tools
• Does not guarantee availability or accuracy

What happens when you submit a message

When you submit a message, Summarly sends the content to Anthropic's Claude AI API. The AI analyses the content and returns a structured explanation. That response is then formatted and displayed to you.

Submitted content is sent to Anthropic under their API terms. By default, Anthropic does not use API inputs to train their models. You can find Anthropic's privacy and usage policies at anthropic.com.

Summarly does not receive, store or retain the content you submit beyond the moment of processing. The content is used only to generate your result and is then discarded.

What is not stored

Summarly does not store:

• The text, images or QR codes you submit
• The results returned to you
• Your name, email address or any identifying personal information (no account is required)

Summarly does log anonymous technical metadata about each session — such as the type of result, response timing, and basic routing information. This does not include the content of your submission and cannot be used to identify you.

Why AI can make mistakes

Summarly uses AI to generate explanations. AI can:

• Miss context that is not visible in the submitted content
• Misinterpret ambiguous or unusual messages
• Produce incomplete or occasionally inaccurate analysis
• Flag a genuine message as suspicious, or miss warning signs in a sophisticated scam

This is why Summarly results should be treated as a starting point. For anything important — a payment, a legal matter, a financial decision, or a genuine threat — verify independently through official channels.

When not to use Summarly

• In an emergency — call 999
• If you believe you have already been defrauded — contact your bank and Action Fraud immediately
• If you need legal, financial, medical or professional advice — consult a qualified professional
• If you are making a compliance or fraud-detection decision for a business — use proper professional processes

When to contact official sources directly

Some situations call for immediate direct contact, not a tool like Summarly:

• Bank fraud or unusual transactions: Call the number on the back of your card or use your official banking app
• Suspected fraud: Report to Action Fraud at actionfraud.police.uk or 0300 123 2040
• HMRC concerns: Contact HMRC directly through GOV.UK or the official HMRC helpline
• Threats or blackmail: Contact your local police or report to Action Fraud
• Identity theft: Contact the affected organisations and report to Action Fraud

Frequently asked questions