DeepHelp Guardian can analyze suspicious message text, submitted URL strings, PDF, UTF-8 TXT and DOCX documents, and PNG, JPEG, or non-animated WebP screenshots. File uploads are limited to 10 MiB.
Frequently asked questions
Clear answers about supported scans, privacy, limits, and safer next actions.
No. The URL scanner analyzes only the characters and locally visible structure of the submitted URL string. It does not fetch the destination, follow redirects, render the website, verify the sender, or query DNS.
No. DeepHelp provides decision support based on available evidence and confidence. A low-risk result is not a guarantee of safety, and file analysis is not antivirus or malware detection.
Raw uploaded files and full extracted text are processed in memory and are not written to a database, object storage, or temporary files by DeepHelp AI. History records retain a content hash, a generated preview, metadata, and the verdict.
Supported uploads are PDF, UTF-8 TXT, DOCX, PNG, JPEG, and non-animated WebP files up to 10 MiB. Documents and screenshots are analyzed for content risk; they are not scanned for malware.
DeepHelp extracts text locally and in memory first. If local extraction is insufficient, the file may be sent to our AI provider as a fallback with request storage disabled and without tools or browsing. Provider extraction can supply text for analysis, but it cannot assign or change the risk verdict. Raw uploads and full extracted text are not retained by DeepHelp after processing.
The public beta uses rolling browser-session and network limits, short-window route limits, and concurrency controls to protect reliability and manage provider capacity. When a limit is reached, the interface shows when to try again.
No. Waitlist signup is not identity verification and does not change the current anonymous beta limits.
Ask DeepHelp provides text-only guidance about scams, phishing, suspicious messages, links, payment requests, impersonation, AI-generated deception, account and identity safety, DeepHelp results, and DeepHelp AI features, privacy, limitations, and supported scans. It does not browse, fetch URLs, take external actions, or assign scan scores or verdicts. Its answers are guidance, not completed scans or guarantees of safety.
Pause before clicking, paying, replying, downloading, or sharing information. Verify the request through a trusted channel you find independently. If money, an account, or personal safety may be affected, contact the relevant bank, platform, authority, security professional, or emergency service directly.
No. DeepHelp is decision-support software. Contact the appropriate bank, platform, authority, qualified security professional, or emergency service when the situation requires official help or urgent action.