// Legal

Privacy Policy

This site sets no analytics cookies and runs no third-party analytics, advertising, or session-replay scripts. We measure page use with our own limited analytics. Here is what we collect, why, and how long we keep it.

Effective
October 8, 2026
Entity
Taint Labs Corporation

01Scope

Taint Labs Corporation publishes adversarial evaluation results for AI agents and is building Stryct Studio. This is our notice, as controller, for the personal information we handle.

It covers two public sites — stryct.ai, including the leaderboard, the platform page, the account sign-up and sign-in pages and the vulnerability report page, and the documentation at docs.stryct.ai — plus email you send to a stryct.ai address.

It does not cover private Stryct Studio engagements. If your organization runs evaluations with us under a contract, data inside that engagement is governed by that agreement and its data processing addendum, not by this page. Where the two disagree about engagement data, the agreement wins.

It does not cover other companies' sites. Our documentation cites papers and advisories hosted elsewhere.

02Cookies and similar technologies

We set no cookies of our own for analytics or preferences. Choosing light or dark from the theme control saves one localStorage entry named taint-theme, containing only light or dark. It is not sent to us. If you never use that control, we do not write it. Our analytics writes no cookies, localStorage, or sessionStorage entries. We use no third-party pixels, advertising tags, session replay, or tag manager.

One exception, and it is not ours to set. Cloudflare’s bot protection sits in front of these sites, and when it screens a request it may place a short-lived __cf_bm cookie in your browser, or a cf_clearance cookie if you were shown a challenge. They exist to tell an automated client from a person. They carry no analytics, feed nothing we read, and are not used to recognize you or to build a profile.

Sign-in cookies. The account pages — sign-up, sign-in, and the page you land on afterwards — use Clerk, our authentication provider. Once you create an account or sign in, Clerk sets the cookies that keep you signed in: session cookies on stryct.ai (such as __session and __client_uat) and a cookie on its own sign-in host, clerk.taint.ai (such as __client). They are strictly necessary to sign you in, carry no analytics, and are not used for advertising. Your browser sends the stryct.ai ones with later requests to the site; we do not read them anywhere except to tell whether you are signed in. If you never use the account pages, Clerk sets nothing.

Apart from Clerk’s script on those account pages, there are no third-party scripts on our pages: no analytics vendor, advertising or retargeting tags, social widgets that phone home, chat widget, session replay, heatmap recorder, CRM tracker, or A/B testing service. Our own analytics script sends limited events to our Worker, as described in section 05.

We currently show no analytics consent banner because this measurement uses no cookies or browser storage and is limited to our own sites. Privacy and electronic-communications rules differ by location. We will review this approach as the rules and our practices change; a missing banner is not a promise that every possible kind of analytics needs no consent.

Do Not Track and Global Privacy Control. If your browser sends Do Not Track or Global Privacy Control, we still count page views but do not create an analytics visitor hash or a cross-view session identifier. A random per-view identifier may still pair a view with its exit event; it is not used to recognize a later page view. We do not sell or share information for cross-context behavioral advertising, with or without either signal.

03What we collect, at a glance

Information reaches us through account sign-up, the vulnerability report form, our first-party website analytics, email you send us, and demo bookings. Each has its own section below; this table is the short version.

CategoryExamplesWhy we have itRetention
Account and waitlist details Email address, your name if you give one, and how you sign in — a password, which Clerk stores as a hash and we never see, or the account of a sign-in provider you choose; plus Clerk’s session and sign-in security records, which include IP address and browser user-agent Signing you in; contacting you when Stryct Studio access opens to you Until access is resolved or you delete your account
Report details Everything you type into the vulnerability report form — title, severity and your reasoning for it, description, reproduction steps, references, and the fields specific to the track you chose — plus your email address, your name if you give one, your credit preference, and the IP address, country and user-agent of the submission Investigating and keeping a record of a security report; replying to you No fixed period: while the matter is live, then our business records
Anti-abuse records A SHA-256 hash of the email address on a report, in a send counter Stopping the report form being used to mail someone who did not ask for it 30 days after it is written
Website analytics events Page path without query string, event type, observed visible time and scroll depth, leaderboard model-row openings and observed dwell, clicks on links, buttons, leaderboard rows and expandable findings (the order of the click on the page, a fixed name for the kind of element, the destination reduced as described in section 05, and the fixed identifier of the row or finding), referring website hostname, selected campaign tags, country, coarse browser, operating system and device type, a per-view identifier, and a daily changing visitor hash and session identifier unless Do Not Track or Global Privacy Control is set Understanding traffic and which pages and results are useful Daily deletion once events are older than 30 days
Daily analytics totals Counts by page, country, referring website, campaign, browser/device class, and named interaction; no IP address, visitor hash, session identifier or per-view identifier Comparing daily interest without retaining individual events Daily deletion after the 30-day reporting window
Correspondence Whatever your email contains, plus its headers; the details you give when you book a demo Answering you; keeping a record As long as the matter is live, then our business records

What we do not collect

Some categories are absent. No passwords reach us: Clerk holds them, hashed. No payments, so no card or bank details. No comments, forums, or uploads; the report form takes text only. Analytics never records what you type into a form, full referring URLs, raw IP addresses, or raw user-agent strings. Sign-in security and the report form are separate and still handle IP addresses as described in sections 04 and 08. We do not collect biometric, genetic, or health data through these sites, government identifiers, or precise GPS location. We perform no automated decision-making with legal or similarly significant effects.

04Accounts and the waitlist

Stryct Studio is not open yet. When you create an account on the sign-up page, your account goes on the waitlist, and the page you land on afterwards says so.

Accounts are run by Clerk, Inc., our authentication provider. Clerk receives what you enter to sign up — your email address, your name if you give one, and a password or the account of a sign-in provider you choose — and stores it on our behalf. It stores a password only as a hash; we never see it. To keep accounts secure, Clerk also records the IP address and browser user-agent of each sign-in session, and may screen a sign-up for automated abuse with a challenge run through Cloudflare. Clerk’s script loads only on the account pages.

When an account is created or changed, Clerk sends our Worker a signed notice, and we copy the account’s email address, name, and Clerk account identifier into a Cloudflare D1 database. That copy is the waitlist.

We use it for one thing: to contact you about Stryct Studio access. We do not add you to a marketing list, send unrelated newsletters, or pass your address to anyone else for their own use. When we do email you, the message will carry a one-click unsubscribe and our postal address, as the law requires. Unsubscribing removes you from the list entirely — there is only the one list. Clerk also sends the emails sign-in itself needs, such as a verification code; those are not marketing.

You can ask us to delete your account at any time by emailing privacy@taint.ai, with no reason required. Deleting the account at Clerk removes your waitlist entry automatically. Deleting it means we can no longer tell you when access opens.

05First-party website analytics

Our Cloudflare Worker records a page view when it serves a public HTML page. A small script on that page sends limited interaction events (including which of the four governance chapters you viewed) back to the Worker using a short-lived, signed page token held only in memory. We store analytics in a dedicated Cloudflare D1 database, separate from the waitlist and report records. We use no outside analytics service, and the script does not set cookies or save an analytics identifier in browser storage.

The script reports visible time and maximum scroll depth when you leave or navigate away; time while the tab is hidden is excluded. On the leaderboard it reports when you open a model or agent row and an estimate of how long that row remained open and visible. Clicks on links, buttons, leaderboard rows and expandable findings produce click events, numbered in the order they happened on that page. Closing a tab may prevent a final beacon from arriving, so measured reading time reflects only visits for which we receive it. This lets us see which published results people inspect; it does not capture the contents of a form, a search, or a private Stryct Studio evaluation.

For an analytics event we retain only:

  • the event type and page path, without a query string or fragment;
  • visible time on the page, capped at 30 minutes, and scroll depth as a percentage;
  • a fixed identifier for an opened leaderboard model or agent row or a tagged call to action, plus observed visible time for an open row;
  • for a click, its order on the page, a fixed name for the kind of element clicked (for example a footer link or a finding being expanded), the fixed identifier of the leaderboard row or finding it belongs to, and its destination: a page on our own sites as its path without a query string, another website as its hostname only, and an email link only when it is one of our published addresses;
  • the hostname of the website that referred you, not the full referring URL, plus the campaign parameters utm_source, utm_medium and utm_campaign if present;
  • country, browser family, operating-system family and mobile or desktop class;
  • a random identifier for each page view, held in memory to pair opening and leaving events; plus a short visitor hash that changes every UTC day and a random session identifier reused for that visitor's events within a 30-minute window. If your browser sends Do Not Track or Global Privacy Control, we omit both the visitor hash and session identifier.

The Worker briefly uses the request IP address and browser user-agent string with a secret key and the UTC date to derive the visitor hash, then discards those inputs for analytics. The key stays on the server; the date changes the hash each day. Neither the raw IP address nor the raw user-agent string is written to the analytics database or our application logs. The hash and session identifier are intended for daily counts and short visits, not for following visits across days. Both are pseudonymous identifiers while retained, and we treat them as personal information where applicable. The per-view identifier does not persist in your browser between pages. IP addresses used for sign-in or report-form security are separate from analytics, as described elsewhere in this policy.

We use these measurements to understand daily traffic, which documentation and results are read, and which links people follow. Automated requests are filtered before analytics are saved. We do not use the data for advertising or to follow people across sites.

06Daily aggregate statistics

A scheduled job combines analytics events into counts by day, page, referring website, campaign, country, browser or device class, named interactions, and clicks by element and destination. These totals contain no visitor hash, session identifier, per-view identifier, IP address, or user-agent string. We no longer email individual visit records or an analytics digest. Copies of digests delivered under the previous system may remain in a company mailbox until deleted under our records process. Both new analytics events and daily totals are deleted after 30 days.

07Email you send us

If you write to legal@, privacy@, security@, demo@, or press@ stryct.ai, we receive whatever is in that message: your address, your name if you sign it, any attachment, and the routing headers your mail server adds.

We use it to answer you and keep a record; security reports in particular are kept so we can track a finding to resolution. Do not send us anything you would not want written down, and never send credentials in plain email.

Booking a demo

The “Book a demo” buttons take you to a Google Calendar booking page belonging to one of our team, taking turns between them. Our Worker keeps a single running count to decide whose page is next; it stores nothing about you or your click. Once you are on the booking page you are on Google’s site, under the Google privacy policy. The name, email address and anything else you enter there, and the calendar invitation that follows, are held in that team member’s Google account. We use them to hold the meeting and follow up on it, and keep them like correspondence.

08Vulnerability and research reports

The report form takes two kinds of submission: a flaw in something we run, and an agentic attack technique you found in somebody else’s model, SDK, or agent. Both write to the same table in our Cloudflare D1 database. It is the only form on this site that takes free text, so it is worth setting out in full.

What the form stores

Everything you typed, as you typed it, plus a little about where it came from:

  • the reference we generate for the report, the TL- code you see on screen and in your acknowledgment;
  • which of the two tracks you chose;
  • the title, the severity you picked, your reasoning for that severity, your description of the issue, your reproduction steps, and any references you added;
  • for a report about our systems, the affected URL or endpoint;
  • for a report about somebody else’s agent, the attack class, the model provider, exact model and version, any agent or SDK you named, whether you judged the behavior reachable on the minimal path, how many attempts you made and how many succeeded, and whether the vendor already knows;
  • your email address, which is required, and your name, which is not;
  • whether you want credit or would rather not be named;
  • the IP address the submission came from, the country Cloudflare derives from it, and your browser’s user-agent string.

There is no file upload on that page and we do not want one, so nothing is stored but text. We ask you not to put a live exploit payload, a real credential, or anyone else’s data into the form. If you do it anyway we hold it, which is exactly why we ask.

The two emails

When email delivery is configured, a submission attempts to send two messages through Resend, our transactional email provider. The report is saved even if mail is unavailable; the confirmation page tells you when delivery is not configured.

The first message, when sent, is an internal notification to security@taint.ai. It carries the whole submission — including your address, and the IP address and country it came from — and it stays in a Taint Labs mailbox afterwards.

The second, when sent, goes to the address you gave, and it deliberately contains nothing you typed: not the title, not the description, not even your own name. It is a reference number and a few fixed sentences. Anyone can type anyone else’s address into a public form, and if we echoed the submission back, this form would become a way to put a stranger’s text in front of you in a message signed by our own domain. We are not going to build that.

For the same reason we cap acknowledgments at two per address in any 24 hours. That counter stores a SHA-256 hash of the lowercased address and the time we sent, never the address itself, and a job on our daily schedule deletes rows more than 30 days old. Past the cap only the acknowledgment is suppressed: the report is still stored and, when mail is configured, the internal notification still goes out.

What protects the form

Cloudflare rate-limits submissions at its edge, per IP address and in total; those counters live in Cloudflare’s network and we store nothing from them. The form carries a hidden honeypot field a human never sees, and a submission that fills it is discarded. When the page loads it asks our own Worker for a short signed timestamp. The Worker also checks browser origin headers before accepting a report. These checks reduce cross-site requests and automated abuse; they cannot prove who typed an email address, or stop a script that imitates a browser.

That timestamp is signed, not stored. It sets no cookie, we keep no copy of it, it lives in the page’s memory until you submit, and there is nothing on our side it could later be matched back to you with. There is deliberately no CAPTCHA, because a CAPTCHA means a third-party script on the page you use to file a security report. What section 02 says about cookies and third-party scripts holds on that page exactly as it does on every other.

How long we keep a report

Indefinitely, in the sense that nothing deletes it on a timer. The daily purge jobs that clear analytics and rate-limit rows do not touch the report table. A vulnerability report is a security record: we keep it while the matter is live and then in our ordinary business records, the same treatment as correspondence, because a finding from years ago is sometimes the thing that explains a decision today.

You can still ask. Email privacy@taint.ai and we will tell you what we hold and delete what we can. We may decline to delete the report itself while it is open or where we need it as a record — if we do, we will say so and say why, rather than let the request go quiet.

If a technique you send becomes a catalog entry, what gets published is the technique, and your credit if you asked for it. Not your email address, not your IP address, and not the raw text of what you sent.

10How we share information

We do not sell personal information and we do not share it for cross-context behavioral advertising. We run no advertising at all and have no data-broker or ad-network relationships to disclose. Personal information leaves our control in four situations only.

Processors

These companies process personal information on our instructions, under contract, for no purpose of their own:

  • Cloudflare, Inc. (United States) — hosting and edge delivery, CDN, DDoS and bot protection, the D1 database holding waitlist entries and vulnerability reports, and a separate D1 database holding website analytics. Cloudflare also rate-limits the report form at its edge and processes request logs as part of operating the network. See the Cloudflare privacy policy.
  • Resend (United States) — delivery of the internal notification for each vulnerability report and the acknowledgment we send to whoever filed it. The notification carries the whole submission; the acknowledgment carries nothing but a reference number. We do not send analytics events or visit digests through Resend. See the Resend privacy policy.
  • Clerk, Inc. (United States) — account sign-up and sign-in: it stores your account, sends sign-in emails such as verification codes, keeps session and sign-in security records, and sets the sign-in cookies described in section 02. See the Clerk privacy policy.

Beyond them, the provider that hosts the mailboxes we read stores messages you send us and report notifications in the ordinary way email works, and demo bookings are held in Google Calendar as section 07 describes. If we add a processor, we name it here and move the effective date.

Coordinated disclosure, if you ask for it

The report form lets you tell us that a vendor does not yet know about a technique and that you would like us to coordinate. If you ask for that, we pass on the technical content of the report and nothing that identifies you — not your name, not your email address — unless you tell us to. If you would rather we introduced you, say so and we will. We do not contact a vendor about your report on our own initiative without telling you first.

Legal and compliance

We will disclose information where the law requires it, or in response to valid legal process such as a subpoena, court order, or lawful government demand. We will also disclose where we reasonably need to in order to investigate abuse of our systems, enforce our terms, or protect people's rights and safety. We read demands narrowly, push back on overbroad or defective ones, and tell you about a demand for your data wherever we are legally permitted to.

Business transfers

If Taint Labs Corporation is involved in a merger, acquisition, financing, reorganization, or sale of assets, personal information may transfer as part of that transaction or its diligence, subject to confidentiality protections. Any acquirer stays bound by this policy for information collected under it until you get notice of a different one and a meaningful chance to respond.

11How long we keep things

We keep personal information only as long as it is doing a job:

  • Accounts and waitlist entries are kept until your Stryct Studio access is resolved — you get in, you decline, or the program closes — or until you ask us to delete, whichever is first. Deleting an account at Clerk deletes its waitlist entry with it. Clerk’s session and sign-in security records are kept by Clerk on its own schedule.
  • Analytics events and daily totals are deleted after 30 days by a daily job. The daily visitor hash is no longer retained once its events have been deleted.
  • Cloudflare operational request logs are retained by Cloudflare on its own schedule and expire without our intervention. Our analytics code does not write raw IP addresses or user-agent strings to our application log.
  • Vulnerability and research reports are not deleted on a timer at all — the daily job does not touch them. We keep a report while the matter is live and then in our business records, the same as correspondence, because a security record is one we may need long after it is closed. Section 08 has the detail.
  • Acknowledgment send counters — the hashed addresses described in section 08 — are deleted 30 days after they are written, by the same daily job.
  • Correspondence is kept while the matter is live, then as part of our ordinary business records for as long as we may need it to resolve disputes or meet a legal obligation.

That job runs once a day, which is worth being precise about: a record can outlive its window by up to a day before the job reaches it, and if a run fails it survives until the next one succeeds. A failed run is logged as a failure for exactly that reason. Read the periods above as the window we delete on, not an instant cutoff.

You can ask us to delete sooner. Email privacy@taint.ai and we will, unless a law requires us to keep a particular record — in which case we will tell you which, and why.

A deletion request covers your account at Clerk and the waitlist entry copied from it. Website analytics is deliberately difficult to connect back to a named person: the visitor hash changes daily and is absent entirely for Do Not Track or Global Privacy Control. If you can provide information that lets us locate an event while it still exists, we will address the request; we will not ask for extra personal information merely to search for a short-lived pseudonymous event. Aggregate totals cannot be attributed to one visitor. A vulnerability report is the exception set out in section 08: we delete what we can, and where we keep a report as a security record we say so plainly. Cloudflare’s operational logs age out on Cloudflare’s schedule rather than ours.

Backups and logs cycle on their own schedule, so a deleted record may persist briefly in a backup after it is gone from the live database. It is not restored to active use.

12Security

Traffic to stryct.ai and docs.stryct.ai is served over TLS, so what passes between your browser and us is encrypted in transit. Our sites run on Cloudflare's infrastructure and inherit its network-level protections, including DDoS mitigation.

Access to the waitlist database, the separate analytics database, and vulnerability reports is limited to the small number of people at Taint Labs who need it for their work. We do not email analytics events. When report notifications are configured, each puts a full copy of that submission in a company mailbox; a mailbox is a weaker boundary than the database, so we disclose that here.

We hold no security certification and will not imply one. We are not SOC 2 audited, not ISO 27001 certified, not HIPAA-covered, and have no third-party penetration test attestation to point at. If that matters to your evaluation of us, ask and we will tell you where we stand rather than send you a badge.

No system is perfectly secure. If you find a weakness in our sites, tell us at security@taint.ai before telling anyone else.

If personal information we hold is exposed, we will say so. Where the law requires it, we notify the relevant supervisory authority within 72 hours of becoming aware, and we tell affected people directly where the risk to them is high — by email for anyone on the waitlist, and on this page for visit records, because there we have no way to reach you.

13International transfers

Taint Labs Corporation is a US company and all three processors are US companies. If you visit our sites or write to us from outside the United States, your information is transferred to and processed in the United States, where privacy law differs from your own and where public authorities may in some circumstances compel access to data.

For personal data originating in the EEA, the UK, or Switzerland, transfers are made under the European Commission's Standard Contractual Clauses, the UK International Data Transfer Addendum where applicable, and the Swiss equivalent — in each case as implemented through our agreements with Cloudflare, Resend and Clerk, with the supplementary measures those agreements require.

One transfer is less formal than the rest. Mail sent to us and the internal notification for each vulnerability report are stored in the United States by the provider that hosts our mailboxes, under that provider’s own terms rather than clauses we negotiated.

We do not claim certification under the EU-US, UK, or Swiss-US Data Privacy Framework.

14Your rights in the EEA, UK, and Switzerland

If the GDPR or UK GDPR applies to you, you have the right to:

  • Access the personal data we hold about you and get a copy.
  • Rectify data that is inaccurate or incomplete.
  • Erase your data where the grounds in Article 17 apply.
  • Restrict processing while a dispute about accuracy or legitimate interests is resolved.
  • Object to processing based on legitimate interests, including our website analytics, on grounds relating to your particular situation.
  • Port data you gave us that we process by consent or contract, in a structured, machine-readable format.
  • Withdraw consent at any time, by asking us to remove your waitlist entry or by using the unsubscribe link in any email we send.
  • Complain to your national supervisory authority, or in the UK to the Information Commissioner's Office. We would rather you came to us first, but that is your choice, not a precondition.

To exercise any of these, email privacy@taint.ai and say what you want. We respond within one month, extendable by two further months for a complex or repetitive request — if we extend, we tell you within the first month and explain why.

How we verify a request

We ask for enough to be confident you are who you say you are, and no more. For a waitlist entry, sending the request from the address on it is normally sufficient. Analytics events hold no raw IP address or email address, and their daily changing hash may make an individual record impossible to locate or verify. If we cannot identify an event, we will explain that rather than collect more information about you purely to run a search. We do not charge for requests, though a manifestly unfounded or excessive one may attract a reasonable fee or be refused, with reasons.

15Your rights in California

The CCPA, as amended by the CPRA, reaches businesses above revenue and volume thresholds we are below today. We offer these rights to California residents anyway, because a company that audits other people’s systems should not be picky about its own.

  • Know. Ask what categories and specific pieces of personal information we collected, where we got them, why, and who we disclosed them to.
  • Delete. Ask us to delete personal information we collected from you, subject to statutory exceptions.
  • Correct. Ask us to fix inaccurate personal information.
  • Opt out of sale or sharing. There is nothing to opt out of.
  • Limit the use of sensitive personal information. There is nothing to limit.
  • Non-discrimination. We will not deny you anything, charge you differently, or degrade your experience for exercising a privacy right. We run no financial incentive program.

Sensitive personal information. We collect none of the categories CPRA treats as sensitive: no government identifiers, no account log-in or financial credentials, no precise geolocation, no racial or ethnic origin, religious or philosophical beliefs, or union membership, no message contents other than mail you deliberately send us or a report you deliberately file, and no genetic, biometric, health, or sex-life data. Website analytics stores country, not city or precise location.

How to exercise your rights. Email privacy@taint.ai. We aim to acknowledge within ten business days and to answer within forty-five days, and we will tell you if something needs longer. Verification works as described in the previous section, and where a record cannot be linked to you at all we will say so.

Authorized agents. An authorized agent may submit a request for you. We will ask for written permission signed by you, and may ask you to confirm it directly and verify your own identity, unless the agent holds a valid power of attorney.

16California categories of personal information

This maps what we collect to the statutory categories in California Civil Code section 1798.140(v)(1), covering the preceding twelve months. We publish it whether or not the statute reaches us. Each category we collect is disclosed to the providers named in section 10, for the business purposes described above, and to no one else except as set out under legal and compliance disclosure. None is sold or shared.

Statutory categoryDo we collect it?What it is, for usBusiness purpose
A. Identifiers Yes Account email and name; the name, email address and report reference on a vulnerability report; IP address in sign-in security and report records; per-view analytics identifier, daily changing visitor hash and short-lived session identifier where privacy signals are absent Accounts and waitlist contact; handling security reports; anti-abuse; traffic measurement
B. Customer records (Civ. Code 1798.80(e)) Yes, overlapping Name and email address, where treated as a customer record — from an account or from a vulnerability report Waitlist contact; replying to a security report
C. Protected classifications No — —
D. Commercial information No We sell nothing on this site and hold no purchase history —
E. Biometric information No — —
F. Internet or network activity Yes Page paths viewed, referring website hostname, selected campaign tags, visible time, scroll depth, and named leaderboard, link and button clicks; the user-agent recorded with a vulnerability report Public-site analytics; handling security reports
G. Geolocation data Yes, coarse only Country derived from a website request; the country recorded with a vulnerability report Where interest comes from; handling security reports
H. Sensory information No — —
I. Professional or employment information Only if you send it Whatever appears in an email you write to us, such as a signature block, or in the free text of a vulnerability report, such as an affiliation or a job title Answering your message; handling your report
J. Education information No — —
K. Inferences and profiles No We do not infer a city or build a visitor profile from website analytics —
Sensitive personal information No None of the categories in section 1798.140(ae) —

17Your rights in other US states

Twenty states now have comprehensive privacy laws. Nearly all of them apply only above traffic or revenue thresholds we are well below, so most do not reach us today. That is not a good reason to give you fewer rights than a Californian. Wherever you live in the United States, the rights in the two sections above are yours: know, delete, correct, and be treated no differently for asking. Email privacy@taint.ai.

If we turn a request down, you can appeal. Reply to our answer and say you want it reconsidered. We will look at it again and write back within forty-five days with the outcome and our reasoning. If the answer is still no, several states let you complain to your attorney general, and we will tell you how.

18Children

These sites are for security engineers, researchers, and the people who buy their tools. They are not directed to anyone under 16, and we do not knowingly collect personal information from anyone under that age.

If you believe a child has given us personal information, write to privacy@taint.ai and we will delete it. We do not sell or share the personal information of minors, because we do not sell or share personal information at all.

19Changes to this policy

We update this policy when our practices change — a new processor, a new data flow, a change to what the analytics script sends — and move the effective date at the top of the page when we do. That date always reflects the published version.

For a change that materially reduces your protections or expands what we do with information already collected, we will post a notice on this page for at least thirty days before it takes effect, and say plainly what changed. Where we relied on your consent, we will ask again rather than assume it carries over.

We will not quietly add cookies or third-party trackers. If that ever changes, it gets announced, not buried in a diff.

Revisions

  • October 8, 2026 — replaced the waitlist form with account sign-up through Clerk, added Clerk as a processor, and described the sign-in cookies it sets on the account pages and the copy of each account that forms the waitlist. The rate-limit table the old form wrote to is no longer used.
  • October 7, 2026 — the analytics script now records clicks on links, buttons, leaderboard rows and expandable findings, in page order, with the destination reduced to a path on our own sites or a hostname elsewhere. Same database, same 30-day deletion, still no cookies or browser storage.
  • September 29, 2026 — replaced the previous page-view log and emailed visit digest with first-party interaction analytics in a separate database. The new system records named page and leaderboard-row interactions, keeps raw events and daily totals for 30 days, does not retain raw IP addresses, full referring URLs or browser user-agent strings for analytics, and suppresses the daily visitor hash and session identifier when Do Not Track or Global Privacy Control is set. Waitlist and report-form security records remain separate.
  • September 11, 2026 — added section 08 for the vulnerability and research report form, with the fields it stores, the two emails it sends, the hashed acknowledgment counter, and the retention, sharing and California entries that follow from it. Later sections renumbered.
  • September 10, 2026 — first published.

20Contact

Privacy questions, rights requests, and complaints go to privacy@taint.ai. A real person reads that address.

Other ways to reach us are listed on the contact page. Other addresses, if they fit better: legal@taint.ai for contracts and legal notices, security@taint.ai for vulnerability reports, demo@taint.ai for Stryct Studio, press@taint.ai for press.

The controller of the personal information described here is Taint Labs Corporation. Our postal address is available on request — email privacy@taint.ai and we will provide it. Formal legal notice goes to legal@taint.ai, as set out in our terms.