Seal the screenshot before the tweet gets deleted.
A politician posts something, you screenshot, you seal in five seconds. The post gets deleted in an hour. The receipt is now dated, signed, and externally anchored — and that timestamp predates the deletion by mathematics rather than your word. Plus: for confidential drafts shared with sources, editors, or outside counsel, mint one invisibly-watermarked copy per recipient; if one leaks, identify which copy escaped.
Workflows journalists actually run
- 01
A public figure posts something and the post disappears.
Seal the screenshot the moment you take it. Even seconds count when you're racing a delete. Once sealed, the timestamp is anchored to a public blockchain — even we cannot retroactively change it. The receipt URL is shareable as an embed-friendly proof: anyone can click through, verify the SHA-256 + the OpenTimestamps anchor offline, and confirm the image existed at that moment without trusting you or us.
- 02
Pre-publication evidence holds.
Seal every working screenshot in your investigation, daily. Five months later when the story drops, the receipts are the timeline of what you knew, when. The verdict against a 'you fabricated this after the fact' attack is mathematical, not editorial.
- 03
Sharing a confidential draft with three editors and outside counsel.
Mint four invisibly-watermarked copies via /track, one per recipient. Each copy looks pixel-identical to the original. If one shows up somewhere it shouldn't, drop the leaked file into the Snitch Tracker recovery step — the per-recipient identifier extracts even after the leaker re-screenshot it, posted it to a Reddit thread that ran through their JPEG pipeline, and someone reposted from there.
- 04
Verifying an inbound tip.
A source sends you a 'screenshot' of an internal Slack message. Drop the file into /verify. If it carries a receipts.you QR, you immediately know whether it's the byte-identical file someone sealed earlier (strongest signal), a platform-recompressed copy of that sealed file (perceptually within band), or — most usefully — a paste-the-QR-on-a-fake image (red verdict, paste attack detected). This is the counter-evidence flow: confirming theirs doesn't match a real sealed receipt, not claiming yours is true.
What it maps to in your world
- SPJ Code of Ethics: 'Verify information before releasing it.' Cryptographic byte-identity is the strongest verification primitive available for screenshot evidence.
- Nieman Lab / Reuters Institute working norms on source protection: the receipt URL is a public dated artifact pointing to a 32-byte hash — the source's file never leaves them, the source's identity never reaches us.
- Reporters Without Borders / Freedom of the Press Foundation tooling stack: receipts.you slots in alongside SecureDrop / Signal-as-document for chain-of-custody preservation.
- The Automattic / Wordpress P2 watermark case (covered by 404 Media): documents how invisible watermarking is being used by companies to identify leakers. Snitch Tracker is the receiver-side counterpart — identify which named recipient's variant matches a leaked file, or confirm that a leak doesn't match any of the variants your source held.
Questions this page answers
- “how to prove a deleted tweet existed”
- “screenshot fake news detection”
- “AI generated screenshot”
- “find who leaked my screenshot”
- “invisible watermark journalism”
- “source protection screenshot”
- “newsroom leak detection”
- “tweet deleted before I screenshotted”
- “screenshot timestamp proof”
- “archive.today alternative cryptographic”
Specific answers
How is this different from archive.today / the Wayback Machine?
Archive services capture a snapshot of a URL at a moment. They're great for pages but depend on the archive service continuing to exist. receipts.you captures the cryptographic proof of an image you already have — independent of whether any third-party service survives. The OpenTimestamps anchor remains verifiable forever using public blockchain headers, even if receipts.you disappears tomorrow.
What about a deepfake or AI-generated screenshot someone hands me?
Sealing it doesn't validate truth — anyone can seal anything; we attest to existence at a moment, not to authenticity of content. But when YOU are the one sealing the original screenshot in real time, and a faked variant shows up later claiming to be the real version, /verify can prove the fake doesn't match your sealed bytes. The 'qr_pasted' verdict catches the most common attack: pasting a real QR onto a fabricated image.
Does the Snitch Tracker watermark survive social media reposts?
The DWT+DCT+SVD layer survives JPEG re-encoding, resize from 0.5× to 2×, and crops up to ~50% from any edge. So Instagram / WhatsApp / Twitter / Telegram round-trips: yes. A leaker who deliberately crops out most of the image, rotates it, or runs it through a removal pipeline: no. Honest scope is documented on /track and tested live at /lab.
Can my source seal something without me?
Yes — the source seals locally in their browser, sends you the file plus the receipt URL through your normal secure channel (Signal, SecureDrop, encrypted email). The source never has to interact with you or us to mint the receipt.
Is the source code public?
Yes. Verifier client, watermark algorithm, test rig — all open. The signing key is private. Read the code, run the test rig at /lab, confirm the claims hold on your own hardware before you trust them.
Pricing for a newsroom?
Free at current usage. Per-seat pricing for newsroom batch sealing, REST API with bearer auth, and private receipts is in development. Email [email protected] with your newsroom's volume and we'll figure something out manually until the public tiers ship.
Carry receipts on every screenshot. They're free.
The next deleted post, the next disputed DM, the next leaked draft — the receipt either exists when you need it, or it doesn't. Sealing takes five seconds.