Frequently asked questions

TIP Protocol (Trust Identity Protocol) is an open, patent-pending standard developed by The AI Lab. It helps you verify who created a piece of content and how it was made.

It brings more transparency to online content by clearly showing whether something is:

• Human created
• AI assisted
• AI generated

TIP works across platforms and can be used by anyone.

You can get your TIP ID at vp.theailab.org/get-verified, the official verification portal by The AI Lab.

Be careful: if any other site asks for TIP ID payment or registration fees, it is not from us.

Before verification starts, a consent screen opens with two tabs:

• Terms of Service. How you can use TIP and what The AI Lab agrees to do for you.
• Privacy Policy. What data we handle and what stays on your device.

Both documents are embedded right inside the screen, so you can scroll and read each one without losing your place or opening a new tab. Switch between the two by clicking the tabs at the top. Once you've read them, tick the box and tap Continue to verification.

Your acceptance is stored on this device only, so we don't ask again on every visit. If the Terms materially change in the future, you'll be asked to re-accept once.

Yes. You must be at least 18 years old, or the age of majority in the country where you live (some countries set this at 19 or 21).

The age requirement exists for two reasons:

• Legal capacity. Identity verification involves a binding agreement with The AI Lab through the Terms of Service, and contracts of that kind require legal capacity.
• The ID itself. The government ID you submit (passport, driving licence, or national ID) is itself an adult-issued document in most countries.

If you are under the age of majority where you live, please wait until you reach it before applying. There is no parental-consent pathway because a TIP ID is bound to a single human and signs that person's own content. A parent cannot sign for a minor here.

The age statement is folded into the consent checkbox you tick before verification begins, so you are explicitly confirming it each time the Terms materially change.

A TIP ID is completely free for individuals, journalists, educators, nonprofits, and governments. You verify your identity once, and your TIP ID works for life on every platform that supports TIP.

Think of it like an email address: free to get, yours forever, works everywhere.

TIP works on Chrome, Edge, Brave, Arc, Firefox, and Safari. The website will automatically detect your browser and show the right install steps.

Note: the TIP extension works on desktop browsers only, but you can complete the verification on your phone.

Yes. You can complete the entire verification on your phone. Some features (like the browser extension) only work on desktop, but your TIP ID works on both mobile and desktop.

AI Trust ID is the umbrella name for both. TIP-ID and CTID are the two kinds of AI Trust ID:

• AI Trust ID
• └─ TIP-ID : the AI Trust ID for a person
  tip://id/US-a3f8c91b2d4e7021
• └─ CTID : the AI Trust ID for a piece of content
  tip://c/OH-7f2a91bc3d5e4a-a3f8

Think of it this way: your TIP-ID is the AI Trust ID that proves you are real. Every time you sign or label something you make, that piece gets its own CTID, an AI Trust ID for that specific item. One TIP-ID can sign many CTIDs over a lifetime, one per published piece.

"AI Trust ID" is the friendly umbrella you'll see on badges and certificates. "TIP-ID" and "CTID" are the technical names protocol developers use.

If you upload a photo where the ID sits inside a larger background (table, hand, sofa), our reader struggles to extract fields cleanly because OCR runs over the whole image and the document becomes a small fraction of the pixels.

To fix this, the page detects the document boundary on upload using lightweight edge analysis in your browser, then perspective-corrects the photo to a tight rectangle around just the ID. You always see a banner with two buttons:

• Use original reverts to the photo you actually uploaded.
• Adjust crop opens the manual cropper with the detected box pre-positioned, so you can fine-tune.

If we can’t find the document confidently, we leave the photo alone and show a hint suggesting a tighter crop. If the server-side OCR still struggles after upload, the page surfaces the same hint based on how many fields actually came back.

Yes. Liveness analysis runs entirely in your browser. The camera feed is processed locally in real time. No video is ever transmitted.

Only a single captured frame is included with the final application, used once for face matching against your government ID. We do not retain it after verification completes.

The liveness check runs multiple independent anti-spoofing layers in parallel. Any one failing fails the check.

• Texture analysis (MiniFASNet). A neural net trained to spot phone screens, photos, and printed prints. It runs on three frames during the scan.
• Moiré FFT. A camera filming a screen produces a beat pattern between sensor and screen pixel grids. We detect that.
• Active screen flash. The page flashes a random sequence of colors and measures the reflected light on your forehead. A pre-recorded video cannot react in real time.
• 3D structure. The face mesh tracks 468 landmarks with depth coordinates. A flat replay collapses depth variance; a real face shows clear nose-to-ear depth.
• Parallax during head turn. Real heads move differently from the background. A flat replay shifts uniformly.
• Random challenge order. Blink, turn left, turn right are issued in randomized sequence so a recorded video can’t pre-satisfy them.
• rPPG (heartbeat). Subtle skin-color variations from blood flow are sampled from your forehead. Recorded video tends to lose this signal.

Browsers do not expose iPhone TrueDepth or Android time-of-flight sensors directly. The above software-based 3D verification is what major banks and identity vendors use for the same reason.

No, we do not store your biometric data. We only use a face scan during verification to match it with your ID. Once the verification is complete, it is not saved.

You can recover access on a new device by importing your encrypted .tip.json backup at /signin and unlocking it with the date of birth on your verified ID.

If you have not made a backup, you can also re-issue with the same identity through the standard verification flow on the new device. Your verification history carries forward.

Individual. The Begin Verification button on vp.theailab.org/get-verified is for every human, including creators, journalists, photographers, podcasters, freelancers, students, and anyone who writes or shares online. You get a personal TIP ID that you sign your own work with. It is free and one-time.

Verified Publisher is a separate review reserved for organizations: news outlets, publications, agencies, and multi-author teams that publish content under an organizational name. To apply for that, go to theailab.org/contact and choose Apply to become Verified Publisher as the enquiry type.

If you write under your own name and represent yourself, you are an Individual. If you represent an organization that publishes under an organizational identity, that organization applies separately. The two can coexist: a journalist working for a Verified Publisher news outlet still gets their own personal TIP ID.

If you still have your .tip.json backup file, you do not need recovery. Sign in at /signin, choose Import key file, and unlock it with the date of birth from your verified ID. Use the recovery flow only when the key is genuinely gone.

The full recovery flow is at /recover. Three steps:

1. Re-verify your identity. Same government ID scan and liveness check as first-time verification. Your ID number and date of birth never leave your device.
2. The network recognises you. Your identity inputs generate the same zero-knowledge proof as before. The network finds your existing TIP-ID and offers recovery instead of creating a duplicate.
3. Bind the new key. You confirm. The network rotates your TIP-ID's public key to the new one generated on this device. From that moment, only this device can sign on your behalf.

What stays intact: your TIP-ID, trust score, tier, content history, and every CTID you've issued.

What changes: the old key is permanently invalidated on the network. All previous .tip.json backups become unusable. Back up the new key immediately once recovery completes.

If you need access urgently and the recovery flow is unavailable, contact support at [email protected].

Passport, driving licence, and national ID card. Which types are available depends on what your country issues. Picking your country shows which document types apply.

Yes. Photos taken on iPhone are usually saved as HEIC (Apple's high-efficiency format), and the upload accepts them directly. We convert HEIC to JPEG in your browser before sending the image to our servers, so there's nothing extra you need to do.

If you're on a desktop browser that cannot read HEIC and the upload fails, open the photo in Photos and re-export it as JPG, or upload directly from your iPhone.

Some countries (United States, United Kingdom, Canada, Australia, and others) do not issue a unified national ID card. For those countries, verify with your passport or driving licence instead.

The portal shows a country-specific note explaining which document types apply, with the national flag and a one-line reason. You always have at least one valid path.

Origin codes describe how a piece of content was created. Every CTID carries one. The simplest way to choose the right one is to ask a single question:

Who is the author?

Common scenarios

The AA vs MX trap

The most common confusion is between AA and MX. The line is sharp:

• AA · one author (you). AI is a tool you used along the way to refine, edit, or rephrase what you wrote.
• MX · two authors. Some sections were written entirely by you. Other sections were written entirely by AI. Each section has a different author.

If you removed MX, you would force people into AA or AG when neither is honest. A research paper where the human wrote the methodology and an AI generated the literature review is not AA (you did not write the literature review) and not AG (you wrote the methodology). It is genuinely mixed. That is what MX is for.

The code is part of every CTID and is publicly verifiable. Honest declaration is rewarded across the protocol.

A CTID (Content Trust ID) is the AI Trust ID for a piece of content. It binds a specific item (post, video, image, article) to the TIP-ID of the person who made it, along with an origin code that says how it was made (human, AI-assisted, AI-generated, mixed).

If your TIP-ID is the AI Trust ID that proves you are real, the CTID is the AI Trust ID that proves this specific piece came from you. See Q6 for the umbrella picture.

Anyone can verify a CTID at /verify-record without contacting any central server.

Your TIP-ID is your identity for life, and every time you sign a post, photo, article, or video with it, that piece gets its own CTID so anyone can confirm a real human made it.

Readers see two different experiences depending on whether they have the TIP browser extension installed.

Without the extension, pasting the CTID adds two lines of text to your post:

That plain-text form still carries full verification: anyone can copy the CTID and paste it at /verify-record to confirm it is genuine.

With the TIP extension, readers never see the raw text. The extension detects the CTID automatically and replaces it inline with a compact verification badge showing your authorship and origin code. Both experiences exist in the same post at the same time.

The TIP browser plugin (extension) puts content labelling and verification right where you already work, on the social and publishing sites you use, instead of switching over to the website for every post.

What it does:

• One-click labelling. A Register control appears on the post you are creating or viewing. The plugin reads the page and pre-fills the platform, content type, and post URL, then signs with your connected TIP-ID. No copy-pasting links into a form.
• Badges for your readers. For anyone who also has the plugin, your CTID is shown inline as a compact verification badge (author plus origin code) instead of a raw tip://c/… line. See Q13b.
• Automatic detection. It recognises posts, photos, videos, and articles on supported sites and offers the right content type for you.
• Your key stays local. Connect your TIP-ID once; signing happens on your device with your biometric or signing password, and nothing about your key leaves the browser.

Why install it instead of using Label Content on the website? Both create the exact same verifiable CTID, but the plugin is faster and feels native: you stay on the page you are posting from, the form is pre-filled from the live page, and verification badges render for your readers. The website's Label Content page is always there as a fallback, and it is the only option on mobile, where browser extensions are not supported. For day-to-day labelling on desktop, the plugin is the smoother path.

It works on Chrome, Edge, Brave, Arc, Firefox, and Safari (desktop only). Get it from the Download Browser Plugin button at /get-verified, then connect your TIP-ID.

The trust score is a public number associated with each TIP ID, derived from verification quality, content history, and protocol-defined signals. Scores update over time as the identity remains in good standing.

Yes. The TIP DAG supports a correction CTID that supersedes a previous one. You issue a new CTID referencing the original; verifiers see the latest status when they look up the record.

Staking means putting a few of your own trust-score points on the line whenever you take part in the TIP dispute process. If you're right, you get the points back plus a small bonus. If you're wrong, you forfeit some or all of them. The stake exists for one reason: to make sure people only act in good faith.

Three places you stake:

• Filing a dispute: stake 15 points. Reclaim them plus a +5 bonus if the jury sides with you; lose them if the dispute is dismissed.
• Serving on a jury: stake 10 points. If your vote is with the majority you get them back plus +3. Vote against the majority and you lose all 10.
• Filing an expert appeal: stake 25 points. Reclaim them plus +10 if the appeal is overturned; lose them if it's confirmed.

Staking is also why the TIP dispute system doesn't need to charge users any cash. Your reputation is the currency.

A Reviewer is a verified human who reviews published content and flags potential issues to the community. There's no application form; you become eligible as your trust score grows. Three layers of participation:

• Community Verification (any TIP-ID, any score). You can confirm someone else's content was labelled correctly. The creator gets a small score bump; you earn modest goodwill credit toward your own Reputation Score over time.
• Jury Service (score 600 or higher). You become eligible to be drawn for a small panel that adjudicates content disputes. A score of 600 represents a verified participant with a consistent record of honest labeling. You do not need specialist knowledge; you need a clean history and good judgment. See the jury questions below for how it works.
• Expert Appeal Panel (score 850 or higher). The highest-trust tier of reviewer. Expert panels (3 people) hear appeals against jury decisions. Their decision is final.

Reviewers aren't paid in money. They're paid in reputation: correct calls add to their score; incorrect ones cost staked points. Over time the panel naturally fills with people whose track record on the network is strong.

A Jury is a group of verified humans selected to decide the outcome of a reported content dispute. A small panel of TIP-ID holders is randomly drawn from everyone with a trust score of 600 or higher. The 600 threshold means jurors are verified, established participants with a demonstrated history of honest behavior on the protocol. It is intentionally accessible so that jury pools are large, globally diverse, and not dominated by any single group. New or unverified accounts are excluded; consistent, good-faith participants are included.

How the jury works in plain language:

• Each juror stakes 10 trust points to take the case.
• Voting is private. Each juror first submits a sealed vote (the "commit" phase, up to 72 hours), then reveals it (6 hours). This stops jurors from copying each other and stops anyone retaliating against an unpopular vote.
• Majority wins.
• The decision is binding unless either side files an expert appeal within 48 hours.

Outcomes for the people involved:

• Creator wins: dispute dismissed, creator gets a +5 vindication bump to their Reputation Score.
• Creator loses: score penalty applied based on how much they under-disclosed the AI involvement.
• Jurors in the majority: stake back plus +3 bonus.
• Jurors in the minority: lose all 10 staked points.
• Jurors who abstain: lose 1 trust point. Their vote is auto-revealed so the case is not held up.

The jury system is what makes TIP a self-governing protocol rather than a platform where one company makes all the calls. Anyone with a clean track record can serve.

Selection is random from a scored pool. No human at The AI Lab picks jurors.

• Jury panel: when a dispute reaches the jury stage, the protocol draws a small panel at random from every TIP-ID holder with a trust score of 600 or higher at that moment.
• Expert Appeal Panel: drawn at random from TIP-ID holders with a trust score of 850 or higher.

Conflict of interest: if you are summoned and you have a personal connection to the creator, the disputed content, or a direct stake in the outcome, you should recuse yourself before submitting a vote. Recusal is a first-class action on the jury screen. It is permanently recorded on the TIP DAG and does not forfeit your staked points. You keep them in full.

The sealed-vote design (commit first, reveal later) also limits coordination problems. No juror can see how others voted until the reveal window opens, so there is no visible majority to follow and no unpopular voter to identify and pressure in advance. Impartiality is structurally enforced, not just asked for.

A dispute challenges the declared origin code on a CTID. Any TIP-ID holder can file one. The flow has up to seven stages:

1. File the dispute. Stake 15 trust points. Describe why the origin code is wrong and attach any supporting evidence. The dispute is permanently recorded on the TIP DAG at this moment.
2. Automated first pass. The protocol's classifier reviews the dispute immediately. If it can decide with confidence, it resolves the case without escalating to a jury. Most clear-cut cases end here.
3. Jury escalation. If the classifier cannot decide, the dispute is assigned to a human jury. A panel is randomly drawn from the pool of 600+ trust-score holders.
4. Commit phase (up to 72 hours). Each juror submits a sealed vote. The vote is hashed and recorded but not revealed yet.
5. Reveal phase (6 hours). Each juror reveals their vote. The earlier commitment ensures no one can change their answer after seeing how others voted.
6. Verdict. Majority wins. The decision is binding immediately. Either side has 48 hours to file an expert appeal.
7. Expert appeal (if filed). A panel of three expert reviewers (trust score 850+, staking 25 points each) hears the appeal using the same commit-reveal process. Their decision is final and cannot be appealed further.

Typical timeline: cases resolved at the classifier stage finish in minutes. Cases that reach a full jury and appeal cycle typically close in 4 to 7 days.

You can file a dispute or check the status of an open case at /disputes.

The reviewer flow depends on which tier you're participating in. All three tiers share one principle: your calls are permanently recorded on the TIP DAG and affect your Reputation Score.

Community Verification (any trust score)

Open any content record at /verify-record. If a Verify button appears, you can review it. Check that the declared origin code (OH / AA / MX / AG) matches what you observe about the content. Confirm or flag it. No staking required. The creator earns a small trust bump from each confirmed verification; you earn goodwill credit toward your own Reputation Score over time.

Jury Service (trust score 600 or higher)

You do not apply. You receive a summons when the protocol draws your TIP-ID for an open dispute case. Accept the case (staking 10 points), then:

1. Read the full dispute record and any attached evidence.
2. Submit your sealed vote within 72 hours. Missing the commit window forfeits your staked points and your vote is not counted.
3. Reveal your vote during the 6-hour reveal window. Missing the reveal also forfeits your stake.
4. See the outcome once all votes are tallied. Majority voters reclaim their stake plus a +3 bonus; minority voters lose all 10 staked points.

If you have a conflict of interest, recuse yourself at any point before voting. Recusal is free of any score penalty.

Expert Appeal (trust score 850 or higher)

If you believe a jury verdict was wrong, you can file an expert appeal within 48 hours of the verdict (staking 25 points). Three expert reviewers are then randomly drawn. Each follows the same commit-reveal process. Their decision is final. If the appeal overturns the verdict, you reclaim your 25 points plus a +10 bonus; if it is confirmed, you forfeit them.

Your full reviewer history across all three tiers is visible at /reviewer-history.

Verdict is a final decision on the content label after the review is complete.

In TIP, once all jurors have cast and revealed their votes, the system counts them and issues the verdict. The verdict tells you exactly what the community decided about the content's origin label: whether it was correct, incorrect, or too close to call. The verdict is binding immediately. Either side can file one expert appeal within 48 hours if they disagree. After the expert panel rules, the decision is final with no further appeal.

For jurors, the verdict also determines whether you voted with the majority or the minority, which affects your trust score.

Upheld means the decision was confirmed and stands as final.

In a TIP dispute, when a jury or appeal panel upholds a verdict, the challenge was correct and the outcome holds. No further changes are made to the content record unless a new appeal is filed within the allowed window.

Quorum is the minimum number of votes needed for a result to count.

In TIP, if not enough jurors cast and reveal their votes before the deadline, the case ends with No Quorum and closes without a verdict. No points are awarded or lost in a no-quorum outcome.

Escrow means your points are held securely by the system until the case is decided.

In TIP, your trust points only go into escrow when you actively accept a jury case. Getting the invitation to serve costs nothing. Your points are only at risk once you say yes. If the case resolves in your favour, your escrowed points are returned plus a bonus. If not, the staked points are forfeited.

TIP uses a two-step voting process called commit and reveal to keep every vote honest and independent.

1. Commit. Every juror privately locks in their vote. No one can see what anyone else chose.
2. Reveal. At the exact same moment, all jurors uncover their votes at once. No one can change their answer after seeing how others voted.

This stops jurors from copying each other and stops anyone from being pressured into switching sides. The same two-step process applies to the expert appeal panel. It makes every vote independent by design, not just by policy.

If a juror chooses to abstain, their vote is auto-revealed immediately so the case is not held up. Abstaining costs 1 trust point.

Yes. If either side disagrees with the jury verdict, they can file an appeal within 48 hours of the verdict.

Three independent experts are brought in. The appeal runs in the same two steps as the original jury vote:

• Appeal Commit. Each expert privately locks in their verdict. No one sees each other's choice yet.
• Appeal Reveal. All three experts reveal their verdict at the same moment.

Majority wins. The decision is final. There is no further appeal after this stage.

Filing an appeal costs 25 trust points. If the appeal succeeds you get them back plus a +10 bonus. If it does not succeed, you forfeit the 25 points.

Yes. Weibo (微博) is a first-class supported platform alongside X/Twitter, Truth Social, Threads, and the rest. Pick Weibo from the platform list when registering content.

Available content types match how Weibo works:

• Weibo. Up to 2,000 characters of text (Weibo raised the limit from 140 in 2016).
• Weibo + image. Text plus an image.
• Weibo + video. Text plus a video URL.
• Weibo thread. A series of connected posts.

Post URLs follow either https://weibo.com/{user_id}/{post_id} or the mobile https://m.weibo.cn/status/{id} shape. Weibo allows post edits with visible edit history (since 2018), so pasting the CTID into the body after registration works straightforwardly.

Long-form Weibo articles (长文) up to ~10,000 characters can be registered using the generic Article type with the article URL.

Yes. WeChat (微信) is supported as a first-class platform. Because WeChat is a multi-product app, four distinct creator-content surfaces are available:

• WeChat Moment. A text post on Moments (朋友圈), up to ~2,000 characters.
• Moment + image. Image plus caption on Moments.
• Channels video. A short video on WeChat Channels (视频号).
• Official Account article. Long-form article published via mp.weixin.qq.com.

Edit policy varies by surface:

• Moments: a very short edit window. If your Moment is already permanent, paste the CTID in a self-comment on the post.
• Channels: edit the description from the Channels admin app.
• Official Account articles: editable any time, paste the CTID in the article body.

Public Official Account permalinks follow https://mp.weixin.qq.com/s/.... Moments and Channels are app-first surfaces; we hash whichever URL or text content the surface exposes.

Yes. Truth Social is a first-class supported platform alongside X/Twitter, Threads, Instagram, Facebook, TikTok, YouTube, LinkedIn, and the rest. Pick Truth Social from the platform list when registering content.

Available content types match how Truth Social actually works (it is built on Mastodon, so the model is familiar):

• Truth. Up to 3,000 characters of text.
• Truth + image. Text caption plus an image.
• Truth + video. Text caption plus a video URL.
• Truth thread. A series of connected Truths.

Post URLs follow the pattern https://truthsocial.com/@username/posts/.... After registration we generate a CTID that you paste into your Truth's body. Truth Social allows post edits with visible edit history, so the paste itself is straightforward.

ReTruths (boosts of someone else’s Truth) are not a content-creation event. You only register a CTID for content you author yourself.

Yes. Mastodon is a first-class supported platform. Pick Mastodon from the platform list when registering content.

Mastodon is a federated, ActivityPub-based microblogging network. Many instances exist (mastodon.social, mastodon.online, mas.to, fosstodon.org, infosec.exchange, hachyderm.io, and many more), and TIP works with all of them because the wire format is identical. Pick the platform once; paste any instance URL when prompted.

Available content types:

• Mastodon post. Up to 500 characters of text. The reference Mastodon limit; some instances allow more.
• Mastodon + image. Text caption plus an image.
• Mastodon + video. Text caption plus a video URL.
• Mastodon thread. A series of connected posts (self-replies).

Post URLs follow the pattern https://{instance}/@username/{post_id}, for example https://mastodon.social/@Gargron/109399894586815521. After registration we generate a CTID that you paste into your post's body. Mastodon 4.0 (October 2022) and later support post edits with visible edit history, so the paste-after-registration flow works the same as on Truth Social.

Boosts (re-shares of someone else's post) are not a content-creation event. You only register a CTID for content you author yourself.

1. Sign in on vp.theailab.org.
2. Open the identity menu in the top-right corner.
3. Choose Connect another device.
4. Authenticate with your signing password or biometric.
5. A QR code appears (it may animate through several frames for ML-DSA-65 keys).
6. On the second device, open the TIP mobile web app or browser extension, choose Scan QR Code, and point the camera at the screen until every frame is captured.

Your private key is encoded in the QR and never sent to any server. Treat the screen like a one-time passcode: only scan it on a device you own.

From the verification portal, click Download Browser Plugin. The portal auto-detects your browser and shows the right install steps with animations. Chrome, Edge, Brave, Arc, Firefox, and Safari are all supported.

On your phone, open the verification portal and tap Install Mobile Web App. The button only appears on iOS and Android. Once installed, the app launches like any other home-screen app and supports labelling content offline.

• Android (Chrome / Edge / Brave): tap Install. The browser shows its native install dialog; tap Add to confirm.
• iOS (Safari): the page shows the three-step Add to Home Screen instructions, because Apple does not expose a programmatic install API. Tap the Share button, choose Add to Home Screen, then Add.

The button hides itself once the app is installed or if you dismiss it for the session. If you ever want to re-trigger the prompt, clear your site data for vp.theailab.org.

Yes. The TIP plugin for WordPress is built for bloggers, news organisations, and multi-author publications running their site on WordPress. Once installed, every post you publish is automatically signed with your TIP-ID and gets a CTID (Content Trust ID) that readers can verify without trusting any single platform.

Where to get it:

• WordPress.org (recommended): wordpress.org/plugins/tip-protocol. Install it directly from your WordPress Admin → Plugins → Add New → search for TIP Protocol.
• Direct download (.zip): on the verification portal, scroll to the For bloggers & publishers card and click Download .zip. Upload via WordPress Admin → Plugins → Add New → Upload Plugin.

Compatibility: WordPress 6.0 or newer. Free and open source. Works on self-hosted WordPress sites and on WordPress.com Business+ plans where third-party plugins are allowed.

Who is this for? If you write a blog, run a news site, or publish long-form content on WordPress, this plugin saves you from having to manually create a CTID for every post on /register-content. The plugin handles signing and CTID registration automatically as part of your publish flow.

Yes. On the Sign In page tap Scan QR code at the top of the no-key flow. Your camera opens with a frame guide. The same single button works for both QR sources:

• From another phone or laptop running TIP: on the signed-in device, open Connect another device from the identity menu and authenticate with your signing password or biometric. Point your phone's camera at the QR.
• From the desktop browser extension: open the TIP extension, click DEVICES → Connect Mobile → Show QR Code. Point your phone's camera at the QR.

Either way the scanner accumulates frames (ML-DSA-65 keys animate in 7 frames) and shows progress dots. Once every frame is captured, the page verifies ownership against the network and asks how to re-secure the key on this device (biometric or signing password).

Your private key is encoded in the QR and never sent to a server. The Use a file instead button inside the modal switches you back to the .tip.json import path if the camera is unavailable. Browsers without the BarcodeDetector API (Firefox, older Safari) automatically load a small jsQR fallback so the scanner works everywhere.

Click the Sign In button on the home page (right under "Download Browser Plugin"), then upload your .tip.json backup file. If it is a v2 encrypted backup, you will be asked for the date of birth on your verified ID to unlock it.

Once unlocked, your TIP ID is connected on this device. You can then use Connect another device to pair further devices.

We are in a public preview window before permanent issuance launches. You can run the full verification flow today; the TIP IDs and CTIDs issued during preview reset on launch day so we can roll out permanent issuance with finalized network parameters.

The slim banner at the top of every page shows the launch countdown. You can join the launch waitlist with one click.

Preview TIP-IDs and CTIDs reset at launch. We notify everyone who joined the launch waitlist, and re-issuance is one click on the same device.

Your verified identity history (liveness, ID match, passkey binding) carries forward; only the issued numbers change.

Your email is stored on our verification server and used only for two things: a quick confirmation right after you sign up so you know you are on the list, and a single launch announcement when permanent TIP-IDs go live. We never share or sell waitlist emails.

You can request removal at any time at [email protected]. After the launch announcement, the waitlist is deleted.

No. TIP's protocol parameters are maintained on-chain, so no single node can change them on its own.

A parameter change is a transaction. It must be proposed by a genuine, validated node or Verification Provider, then accepted and validated by the other nodes in the network, exactly like any other transaction. There is no privileged path.

Once the network accepts the change, every node receives the same values at the same time, so the network never splits into nodes running different rules. No single party can force a parameter change through.

Every review follows four phases within a 48-hour window. Watch the full Reviewer Role explainer video for a plain-language walkthrough.

You qualify if all four conditions are true:

• Trust score 800 or higher. Reviewers shape real outcomes, so a proven track record is required.
• Personal identity. Organizations and publishers cannot be reviewers. Only individual humans qualify.
• Opt-in turned on. Go to Profile → Settings and turn on “I want to be a reviewer.” You are never auto-enrolled.
• Overturn rate under 30%. If too many of your past calls have been overturned by juries, you are paused from new assignments until your rate improves.

You can turn the setting off at any time and stop receiving assignments immediately.

• Dismiss. You believe the AI was wrong. The content looks genuinely human-made. The creator is unaffected and you earn +5 trust score immediately.
• Confirm. You believe the AI was right. The creator’s label does not match the content. The creator is privately notified and given 24 hours to correct the label or push the case to a public jury. If it goes public, you become the formal disputer with 15 points at stake.
• Recuse. You know the creator, have a personal or financial stake in the content, or simply do not feel qualified. A fresh reviewer is assigned automatically. No penalty, no bonus.

If you genuinely cannot tell, use Recuse. That is exactly what it is for.

If you lean toward borderline AI involvement rather than clearly AI-generated, you can Confirm with origin AA (AI-Assisted) or MX (Mixed) instead of AG. The creator then has 24 hours to accept that privately or push the case to a jury. The middle ground is honored.

Use Recuse. If you know the creator, have a personal or financial stake in the content, or simply do not feel qualified to judge it fairly, step aside. Recusing is free, instant, and always the right call. No score penalty applies.

No. Once you sign and submit, the decision is on the chain. It cannot be changed or retracted. Take your time reading the content before clicking.

No. You see their TIP ID only. You are judging the content, not the person. The creator’s legal identity is never shown to you during the review.

No. The decision is recorded on the chain and can be audited by anyone, but your TIP ID is not shown as the assigned reviewer to the creator. Anonymity is built into the design.

No. You decide one thing only: does the content look AI-generated when the creator claimed otherwise? All other types of moderation are outside your role as a Reviewer.

There is no money involved. You earn trust score, which is your reputation across the protocol. A higher trust score qualifies you for Juror and Expert roles, each of which carry their own stakes and recognition.

You have 48 hours to act. After that the system auto-recuses you and assigns a fresh reviewer. There is no direct point loss, but your no-show count goes up. If it becomes a pattern, your assignments are paused until your availability record improves.

Missing one case occasionally is fine. Just do not make it a habit.

Yes. Every assignment you close (Dismiss, Confirm, or Recuse) is recorded on the chain and counted toward a Reviewer badge on your profile, for example “Served 42 times as Reviewer.” The badge appears the first time you close a review and ticks up with every one after. There is no threshold to unlock it.

Go to your TIP ID settings and open the Preferences section. You will see an option labelled “I want to vote.” Toggle it to ON to start receiving review invitations. You can turn it OFF at any time to stop receiving new assignments immediately, with no points lost when you opt out.

Only turn it ON when you have time to respond within 48 hours. Every invitation that expires without a response costs 1 trust point, so opting out during a busy period is always better than letting assignments go unanswered.

As a Reviewer you are not required to side with the AI classifier or with the content creator. The system gives you all available origin options so you can form a fully independent judgment.

A Conservative label is when your honest assessment does not fully agree with either party. For example, the AI classifier flagged content as AI-Generated (AG) and the creator claimed Original Human (OH), but your read is AI-Assisted (AA). You can select AA. This independent call has the same protocol weight as a standard Confirm or Dismiss and is recorded permanently on the TIP DAG under your TIP ID.

Use it when you genuinely believe the truth sits between the two positions. Your reputation is attached to every call you make.

When you submit a Confirm verdict the creator receives a private notification and has 24 hours to respond.

If the creator acknowledges the error and updates their content label:

• The creator loses 10 trust points.
• You receive +3 trust points for an accurate classification.
• The case closes immediately with no jury needed.

If the creator does not respond or disagrees within 24 hours, the case automatically escalates to a Stage 2 Public Jury. You become the formal disputer with 15 trust points at stake. If the jury agrees with you, you earn those back plus a bonus. If the jury sides with the creator, you lose them.

This is why it matters to only Confirm when you are genuinely confident. Your case history is permanently on the chain and visible to anyone on the network.

As long as your Recuse rate stays below 30%, it has no impact on your standing as a Reviewer.

If your Recuse rate rises above 30%, the system automatically stops selecting you as a reviewer. You will not receive new invitations until your rate improves through future participation.

Recuse is designed for genuine conflicts of interest or uncertainty, not for skipping difficult cases. If you find yourself recusing frequently, the better move is to turn the Reviewer toggle OFF in your settings until you are ready to engage actively.

Every verdict you participate in (Confirm, Dismiss, or Recuse) is counted toward a Reviewer recognition badge on your public profile, for example “Served 27 times as Reviewer.” The badge grows with each case and is visible to anyone on the network, building credibility and validity over time.

Beyond the badge, correct calls earn trust points. As your score grows, the protocol automatically invites you to progressively higher roles:

• Juror (600 points): you can be drawn to serve on public dispute panels.
• Expert Reviewer (850 points): you hear final appeals, and your decision is binding with no further appeal.

There is no separate application for either role. Once your score reaches the threshold and you have the corresponding toggle ON in your settings, invitations arrive automatically. Watch the Reviewer Role explainer to see the full path.

The juror process runs over 84 hours in four phases. Watch the Juror Role explainer video for a plain-language walkthrough of every step.

You qualify if all four conditions are true:

• Trust score 600 or higher. Jurors decide real outcomes that permanently update the content record on the TIP DAG, so a meaningful track record is required. A score of 600 means you have verified your identity, maintained honest content labeling, and built enough consistent history for the protocol to trust your judgment on a real case. It is the point at which the community considers you established enough to weigh in on disputes affecting other creators. You do not need to be an expert; you need to have demonstrated good faith. The threshold was set at 600 to keep the eligible pool broad and globally diverse while still filtering out brand-new or low-reputation accounts.
• Personal identity. Organizations cannot serve as jurors. Only individual humans qualify.
• Opt-in turned on. Go to Profile → Settings and turn on “I want to be a juror.” You are never auto-enrolled.
• No direct conflict. You must have no personal or financial connection to the content or the parties involved.

You can turn the setting off at any time and stop receiving assignments immediately.

How to reach 600: complete identity verification, register your first pieces of content with accurate origin labels, and participate in Community Verification reviews over time. Consistent, honest behavior is what moves your score. There is no shortcut and no fee.

• Match. You agree with the creator's origin label. The dispute is groundless. The creator wins and the disputer loses their 15-point stake.
• Mismatch. You agree with the disputer. The creator's label is wrong. If you pick Mismatch you must also select the correct origin (OH, AA, AG, or MX) because the chain updates the content label when Mismatch wins.
• Abstain. You genuinely cannot decide. Abstain is neutral at 0 score impact. Your vote does not count toward the majority result, but it does count toward quorum.

To prevent jurors from copying or influencing each other.

In the Commit phase (72 hours), each juror submits a scrambled version of their vote. Nobody can see what anyone chose. In the Reveal phase (12 hours), jurors submit their real vote and the unscrambling key. The system verifies the reveal matches the original commit and all votes become public at once.

This ensures every juror votes based on their own honest read of the content. No one can wait to see which way the majority is leaning before deciding.

You take a -8 trust score penalty. Only you can reveal your vote because only you have the unscrambling key. No one can do it for you.

If you know in advance you will be unavailable, do not commit at all. The penalty for never committing is only -1. Committing and then going silent is penalised more heavily because it occupies a jury slot and disrupts the panel's quorum math.

There is no money involved. You earn or lose trust score, which is your reputation across the protocol.

For a verdict to count there must be at least 5 total reveals AND at least 3 non-abstain reveals. Both thresholds must be met.

If quorum is not reached, the case auto-escalates to a Stage 3 Expert panel. No score changes apply to any juror when quorum fails.

Only after the reveal phase ends. During the Commit phase all 7 jurors are anonymous to each other and no one can see what anyone chose. After Reveal, everyone's vote is on the chain and publicly viewable, but the system does not show fellow jurors to each other in real time during the voting window.

If you never commit a vote in the 72-hour window you take a -1 trust score penalty (as long as the jury reached quorum without you). Your no-show count also goes up.

If you know you will not be available, this is actually the better path than committing and then missing the reveal, which costs -8. Just do not make no-show a habit. If the pattern continues, your assignments can be paused.

Yes. Every jury reveal you submit is recorded on the chain and counted toward a Juror badge on your profile, for example “Served 17 times as Juror.” The badge appears the first time you reveal a vote and ticks up with every reveal after that. Abstain reveals count. No-show cases do not.

The full process can run up to 9 days (216 hours) from filing to final verdict. Most disputes settle at Stage 2.

You qualify if all conditions are true:

• Trust score 550 or higher. This filters out brand-new or low-reputation accounts from filing disputes.
• Personal identity. Organizations cannot file disputes. Only individual humans qualify.
• Not your own content. You cannot dispute content you created. The system filters this automatically.
• Under the 5-per-30-days cap. You may file at most 5 disputes in any rolling 30-day window. This prevents serial disputing as harassment.

Disputes are about origin labels only. Whether the content is good, bad, offensive, or politically disagreeable is not relevant to the dispute process.

15 trust score points are deducted from your score immediately. This happens the moment you click File Dispute and sign. The stake cannot be reversed or refunded unless you win.

At the same moment, 7 jurors are randomly selected and summoned, and the dispute is permanently recorded on the chain. You cannot withdraw it from this point. The creator is notified and the jury clock starts.

Conservative Label means jurors agreed you were right to dispute (the creator's label was wrong), but they could not reach a majority on what the correct replacement label should be.

The system applies the smallest-penalty correction available as a fair fallback. You get your 15-point filing stake refunded but receive no bonus. It is a partial win: you were right, the content gets corrected, and you break even on points.

Only if you genuinely believe the jury got it wrong and you have specific evidence they clearly missed. The math does not favor speculative appeals.

• Lose at jury and also lose the appeal: -40 points total
• Lose at jury but win the appeal: +15 points net

You have 48 hours after the jury verdict to decide. Do not appeal out of frustration or on a hunch. Appeal only when the evidence is strong.

An appeal costs an additional 25 trust score points staked upfront at the time of filing.

If the jury ruled in your favor and the creator files an appeal to Stage 3, you are a passive participant. You do not need to do anything.

• Experts uphold the jury: you keep your Stage 2 net gain. Nothing changes for you.
• Experts overturn the jury: your Stage 2 win is reversed and you lose your 15-point filing stake. You do not lose 25 additional points because you did not file the appeal.

Only file when you genuinely believe you are right and have evidence. A 50/50 hunch loses points on average.

The expert process runs over 84 hours in four phases. Watch the Expert Reviewer Role explainer video for a plain-language walkthrough. The decision is final with no further appeal.

You qualify if all conditions are true:

• Trust score 850 or higher. Experts hold the final word. The highest trust threshold on the protocol.
• Personal identity. Organizations cannot serve as Expert Reviewers.
• Opt-in turned on. Go to Profile → Settings → Adjudication participation and turn on “I want to help adjudicate.” This is the same toggle as the Juror setting.
• Not on the original Stage 2 jury. You cannot judge the same case twice. The system filters this automatically.
• No conflict of interest. You cannot be the creator, disputer, or appellant in the case.

In one of two ways:

• Filed appeal. After the Stage 2 jury verdict, the losing party has 48 hours to file an appeal at a cost of 25 trust score points. If they file, 3 experts are randomly selected.
• Automatic escalation. If the Stage 2 jury fails quorum (not enough jurors revealed their votes), the case auto-escalates to Stage 3 without anyone needing to file a formal appeal.

Expert panels are rare. Expect a handful of summonses per quarter on an active network.

• Uphold Verdict. You agree with the Stage 2 jury. The jury was right. The Stage 2 outcome stands and the appellant's 25-point filing stake stays forfeited.
• Overturn. You disagree with the jury. The entire Stage 2 settlement reverses. You must also select the correct origin label (OH, AA, AG, or MX) because overturning changes the content label on the chain.
• Abstain. You genuinely cannot decide. Abstain is neutral (0 score impact) but does not count toward the panel majority. If fewer than 2 experts give non-abstain reveals, the appeal defaults to dismissed.

Same reason as the Stage 2 jury: to prevent experts from copying or influencing each other. In the Commit phase (72 hours), each expert submits a scrambled version of their vote that nobody else can read. In the Reveal phase (12 hours), experts submit their real vote and the unscrambling key. All three votes become public at once.

Each expert votes blind and independently. No bandwagoning possible.

Expert penalties are heavier than for Jurors (minority -10 vs -8, missed reveal -10 vs -8) because Experts hold the final word and the trust bar is higher at 850+. The bigger stakes though are the case-level ones: the appellant's 25-point filing fee and the full Stage 2 settlement that reverses on Overturn.

The expert panel needs at least 2 non-abstain reveals (out of 3) to produce a valid verdict.

If fewer than 2 experts give valid non-abstain reveals, the appeal defaults to dismissed: the Stage 2 verdict stands and the appellant's 25-point stake stays forfeited. Experts who did not commit or missed the reveal still take their no-show penalty (-1 for no commit, -10 for commit then bail).

This is different from a Stage 2 jury no-quorum. When Stage 2 fails quorum it escalates to Stage 3. Stage 3 has no further tier, so a failed expert panel defaults to dismissed and the case is closed.

Stage 2 jurors keep their original majority or minority scores from Stage 2. The chain never retroactively rescores them on appeal.

Only the case-level settlement reverses: the creator, disputer, and reviewer (if a prescan review was involved) see their Stage 2 points reversed or re-applied to match the new outcome. Juror scores are sticky by design.

Yes. Every expert reveal you submit is recorded on the chain and counted toward an Expert badge on your profile, for example “Served 4 times as Expert.” The badge appears the first time you reveal a vote and ticks up with every reveal after that.

Because Stage 3 is rare by design, the count grows slowly. That is intentional: the badge signals depth of service, not volume.