← Back to the platform

Privacy Policy

Datenschutzerklärung for the Threatera platform — what personal data the application processes, why, and on what legal basis, under the EU General Data Protection Regulation (GDPR).

Last updated: 11 November 2026 · Applies to the Threatera application. The separate website privacy policy covers the public threatera.eu site.

1. Who is responsible for what

Threatera is a threat intelligence platform. The data it holds falls into two categories with two different legal answers, and telling them apart is the most important thing on this page.

Your account — we are the controller

The data needed to give you a login and keep the service secure — your email address, your password hash, your MFA secret, your role, and the audit record of what you did — is processed by us as controller within the meaning of Art. 4(7) GDPR. Sections 3, 4, 5, 10 and 11 describe it.

What you analyse — we are the processor

The files, URLs, indicators, rules and notes you submit are yours, as is anything that arrives in a reporting mailbox you connect (Section 7). We process them on your instructions, as a processor under Art. 28 GDPR, in order to run the analysis you asked for. We do not decide what you upload, we do not use it for our own purposes, we do not use it to train anything, and we do not sell or share it.

This matters because malware samples contain other people's data. A phishing kit holds the credentials it has harvested. A leaked document is somebody's document. A captured email is somebody's email. When you upload such material you remain the controller of the personal data inside it, and it is you who must have a lawful basis for processing it — typically Art. 6(1)(f) GDPR, the legitimate interest in network and information security that Recital 49 explicitly recognises.

The details of our processor role — scope, instructions, sub-processors, security measures, deletion, and assistance with data subject requests — are set out in the data processing agreement (Art. 28(3) GDPR) concluded with your organisation.

2. Controller

The controller for the account data described in this policy, and the processor for customer content, is:

Philipp Wudernitz
Stockerauer Straße 31a
2100 Korneuburg, Austria

Email: privacy@threatera.eu

We have not appointed a Data Protection Officer. The conditions in Art. 37(1) GDPR are not met: the platform's core activity is the analysis of technical artefacts, not the regular and systematic monitoring of people, and it does not process special categories of data on a large scale as a core activity. Data protection enquiries reach us at the address above.

3. Account data

To give you access to the platform we store:

DataWhy
Email address (used as your username)To identify you at sign-in and to attribute your actions in the audit log
Password, stored only as an Argon2 hashTo verify your sign-in. The password itself is never stored and cannot be recovered from the hash
TOTP secret for multi-factor authentication, which every account that signs in with a password must set upTo verify your authenticator codes
Your role in each organisation you belong to (owner, admin, analyst or read-only), and whether your access there and your account are switched onTo decide what you are allowed to do, and where
Account creation dateAdministration
Where the installation offers self-service trials: the address and organisation name you enter to start one, with a hash of the one-time link we mail youTo prove the address before anything is created. Deleted after seven days whether or not the link was used; if it was, the address becomes your account's username as above

Legal basis: Art. 6(1)(b) GDPR — processing necessary to perform the contract under which the platform is provided to you.

We do not ask for your name, telephone number, postal address, date of birth, or any payment details in the application. Accounts are created by invitation from an administrator in your organisation, not by public self-registration — or, where your organisation has set up single sign-on with automatic provisioning, the first time you sign in through it, in which case the account holds the email address your identity provider asserted.

4. Sign-in, sessions and MFA

When you sign in successfully, the platform issues a signed session token (a JSON Web Token) which your browser stores in localStorage and sends with each request. It expires after 24 hours, after which you sign in again. Signing out deletes it from your browser.

A successful sign-in also returns a known-device token, which your browser keeps in localStorage (under tt_devices, one per account signed in from that browser, at most five) for up to 180 days and sends back the next time you sign in. It holds your account name and a random identifier, and it does one thing: it gives this browser its own allowance of sign-in attempts, so that somebody trying passwords against your account cannot lock you out of it. It grants no access to anything. It is kept when you sign out, because being recognised at your next sign-in is its purpose; changing your password retires it, and clearing this site's data in your browser removes it.

The platform also keeps, in localStorage, which organisation you are working in (tt_tenant), your light/dark theme (tt_theme), table column widths you have dragged (tt_colw), and the width and collapsed state of each tab's side list (keys beginning threatera.rail.). None of it identifies you or is sent anywhere. It sets no cookies of its own, and it runs no analytics, tracking, profiling, advertising or session-recording of any kind.

Multi-factor authentication is based on time-based one-time codes (TOTP), and it is not optional: an account that signs in with a password must enrol an authenticator before it can do anything else. An account that signs in through your organisation's single sign-on relies on your identity provider's own second factor instead. The QR code shown during enrolment is generated inside the platform; it is not fetched from an external service, so your MFA secret is never sent anywhere to be rendered.

Legal basis: Art. 6(1)(b) GDPR (performing the contract) and Art. 6(1)(f) GDPR (our legitimate interest in securing accounts). The browser storage described here is strictly necessary to provide a service you explicitly requested, and therefore requires no consent banner under § 165(3) TKG 2021 / Art. 5(3) ePrivacy Directive.

5. Invitations

An administrator can invite a colleague by email address. We store that address, a single-use invitation token, the intended role, and the expiry and acceptance times, so the invitation can be validated and so an administrator can see and revoke pending invitations.

The invitation email is sent through our mail provider, netcup GmbH (Emmy-Noether-Straße 10, 76131 Karlsruhe, Germany), acting as a processor on our behalf under an Art. 28 GDPR agreement. They necessarily see the recipient address and the content of the invitation. No other party is involved in delivering it. A password-reset link you ask for is sent to your address the same way.

Legal basis: Art. 6(1)(f) GDPR — the legitimate interest of your organisation in provisioning access for its staff.

6. Content you submit for analysis

The platform stores what you give it to work on:

  • Uploaded files — the complete file, kept in object storage and addressed by its SHA-256 hash, together with its size, type and the filename it was uploaded under.
  • Analysis results — antivirus and YARA matches, extracted strings, static structure, capability analysis, and any indicators (domains, IPs, URLs, email addresses, hashes) extracted from the file.
  • Text extracted for searching inside samples — up to 1 MB per file or reported message of the text the platform can read from it: every printable string, document text, macro code, decoded scripts, imports, rule and capability names, what a sandbox run printed, ran and contacted, and a reported message's subject and body. It is kept so that search can find a string inside your samples, per organisation even where another organisation holds the same file, and it is deleted with the file or message by retention and by deleting your organisation. If your organisation holds its own encryption key, this copy is outside that key, so it is not made unless your administrator switches it on, and switching it off deletes it.
  • The contents of archives you upload. When a zip, tar, 7z or rar is analysed, its members are stored as files in their own right so they can be examined and pivoted from, each recorded as having been extracted from the archive it came out of. They are subject to the same access control, retention and deletion as anything you upload directly — deleting your organisation removes them with everything else.
  • Files a sample creates when you detonate it. Up to 25 per run, of 1 MB each and 8 MB in all, are stored as files in their own right, each recorded as created by that run, so they can be opened and analysed. They are subject to the same access control, retention and deletion as anything you upload directly.
  • Indicators you add, with your tags and notes.
  • URLs you submit for sandboxed rendering, and the results described in Section 8.
  • Detection rules (YARA, Sigma) and comments you write, together with the username that wrote them.
  • Investigation records — cases, what is attached to them and why, timeline notes, and, where you record them, the hosts, accounts and systems an incident affected, a business-impact assessment and a post-incident review. A host or account name is often a colleague's laptop or login, so it is personal data: it is shown to members of your organisation and included in the audit pack, and left out of the case report and of the impact text the platform suggests for a regulatory filing, which count hosts and accounts rather than naming them. None of it is included in a STIX export. Each entry records the username that added it.

Uploaded files are kept in full and are not redacted, because an altered sample is not the sample. Please treat every upload as a disclosure: if a file contains personal data, that personal data enters the platform and stays in it until deleted. This applies with particular force to material that may contain special categories of data under Art. 9 GDPR — health records, identity documents, and similar — which is a realistic possibility for exfiltrated data recovered during an incident. Your organisation's own policy on handling such material governs whether it should be uploaded at all.

Legal basis: we act on your instructions as processor (see Section 1). Your organisation is responsible for the lawful basis for the content itself, ordinarily Art. 6(1)(f) GDPR read with Recital 49.

Counts other organisations see — only if your administrator switches it on

Everything above is your organisation's alone, even where another organisation holds the same file. There is one exception, and it is off unless an administrator of your organisation switches it on: counts from other organisations. When it is on, other organisations that have also switched it on can see, for a file or indicator they hold, how many organisations have met it — and you can see the same about theirs.

  • A number, in bands, and nothing else. Fewer than 3, 3–5, 6–10 or more than 10. Never which organisation, never when, never a filename, a verdict or anything found in the file.
  • Only files (by their SHA-256) and indicators (domains and IP addresses). Never URLs or email, which can identify people.
  • Only from the day it is switched on. Nothing your organisation held before then is counted. Switching it off stops your organisation being counted at once.
  • Only about what the asker already holds. No organisation can ask about a file or indicator it does not hold itself.

Legal basis: the administrator's switch is your organisation's instruction to us as processor to include its holdings in these counts (Art. 28(3)(a) GDPR). The setting and every change to it are recorded in your organisation's audit log.

Early warnings between organisations — only if your administrator joins

A second exception, also off unless an administrator of your organisation joins it: early warning. When it is on, values your organisation judged harmful are contributed in a form nobody can read, and when three organisations that joined hold the same value, each of those three is told what it is.

  • What is contributed. From the last 30 days: the SHA-256 of files the platform or one of your analysts called malicious or suspicious, and domains and IP addresses one of your analysts judged malicious or suspicious. Never a file, a filename, a URL, an email or anything found in a file.
  • In what form. A token and a share of a secret, made with a key held by a separate service (the blinder) that the platform's own processes are never given, and the value encrypted under that secret. Below three organisations none of it can be read — not by other organisations and not by the platform's own database.
  • Who is told, and what. Only an organisation that holds the value itself, and only once three do: the value, and how many hold it as a band (3–5, 6–10, more than 10). Never which organisations, never when they contributed.
  • Leaving. Switching it off deletes every contribution of your organisation at once; it then counts toward nobody's warning. A contribution not renewed lapses after 30 days.
  • The blinder. It runs inside the platform's own infrastructure, separately from the platform, and sees only blinded numbers — never a value and never which organisation asked. It makes no connection outside the platform.

Legal basis: the administrator's switch is your organisation's instruction to us as processor to contribute these values and to receive warnings (Art. 28(3)(a) GDPR). The setting and every change to it are recorded in your organisation's audit log.

7. Reported email

Where your organisation switches it on, the platform connects to a mailbox you nominate — ordinarily the phishing@ address your staff already forward suspicious messages to — and analyses what arrives in it: the headers, the body, the links, and any attachments.

This is the most privacy-sensitive part of the platform, and the only part where personal data arrives rather than being chosen. Everything in Section 6 is material an analyst decided to submit. A mailbox holds whatever is sent to it — the sender's name and address, a signature block with a direct line and a job title, a quoted thread naming colleagues and customers. Many of those people are not your staff, never became our customer, and made no decision to be here at all.

It is therefore off by default. No mailbox is polled until someone in your organisation configures one and enables it, and the feature can be disabled for the whole installation (EMAIL_INGEST_ENABLED).

What is stored

  • The complete message, as it was received, kept as a file in its own right like any upload. Nothing is stripped: an analyst reading a phishing report needs the headers exactly as they arrived.
  • The header summary — sender address and display name, recipient, Reply-To, subject, date, and the Message-ID — held as searchable columns so a campaign can be traced across reports.
  • The authentication result the receiving mail server recorded (SPF, DKIM, DMARC), read from the message and never recomputed. Re-checking would query the sending domain's infrastructure, which is both unreliable after the fact and audible to whoever operates it.
  • The findings — why the message was judged as it was: a mismatched Reply-To, a display name claiming a brand the address does not, a link whose text and target disagree, a lookalike sender domain.
  • Attachments, stored as files recorded as having been extracted from the message, so they can be examined and pivoted from exactly as an archive's members are. They are subject to the same access control, retention and deletion as anything uploaded directly.
  • The sender's domain and the host of every link, recorded as indicators so they are searchable and can be matched against the platform's threat feeds.

What is deliberately not done

  • Nothing is deleted from your mailbox. Messages are marked as read and left where they are. The mailbox is yours.
  • No reply is ever sent — not to the reporter, and certainly not to the sender.
  • Links are not opened automatically. A phishing message carries dozens, most of them tracking pixels that would confirm to the sender that the message was read. They are recorded as indicators; opening one is a deliberate act by an analyst, and it then happens as described in Section 8.
  • Nothing is sent to a third party. The only outbound connection is to the mail server you nominated, to read your own mailbox (Section 12).

Credentials

The mailbox password is encrypted at rest and is never displayed again after it is entered. Each organisation's mailbox is separate: one installation-wide mailbox would put one customer's reported phishing, with their staff named in the forwarded thread, into another customer's platform.

If it was your message that was reported

If you wrote to someone whose organisation uses Threatera and they reported your message as suspicious — which happens to legitimate senders regularly — your message and your address are held in that organisation's instance. They are the controller of it, not us. Contact them; if you contact us we will pass the request on without undue delay, as Art. 28(3)(e) GDPR requires (Section 15).

Legal basis: we act on your instructions as processor (Section 1). For your organisation, this is ordinarily Art. 6(1)(f) GDPR read with Recital 49, which recognises network and information security as a legitimate interest. Because this data is not collected from the people it concerns, Art. 14 GDPR applies to your organisation as controller — including the question of whether an exemption under Art. 14(5) is available. It generally is for the sender of a phishing message, where notification would defeat the purpose of the processing; it is a weaker argument for third parties quoted in a forwarded thread. That assessment is yours to make and to document, and it is the reason this feature is not enabled for anyone by default.

8. The sandbox

When you submit a URL, the platform opens it in an isolated headless browser and records what happens. Two consequences are worth stating plainly.

The site you scan sees the platform, not you

The request comes from our infrastructure. The operator of the scanned site learns that the page was fetched, and sees our server's IP address and the sandbox browser's user agent — not yours, and not your organisation's address. That is deliberate. It does mean the site's operator may be able to infer that someone is looking at it.

What the scan stores

  • a screenshot of the rendered page, the page title, and the final URL after redirects;
  • the list of network requests the page made, which becomes the indicators the scan contacted;
  • cookies and browser storage the page set — recorded by name, domain, flags and value length. The values are deliberately not stored, because a session cookie is a live credential and keeping it would put it in our database and in every exported report;
  • whether the page tried to fingerprint the browser, and whether any form on it asks for a password or payment details and where it sends them.

A rendered page can of course contain personal data — a profile, a message, a document behind a link. Where it does, it is stored as part of the scan result and is governed by Section 1.

Files can also be detonated in an isolated environment. Detonation is confined to file types the platform can actually run; where it cannot, the action is disabled rather than silently doing nothing.

A detonated file's network is, by default, a sinkhole: it answers the sample's requests itself and records them, and nothing leaves our infrastructure. An analyst can instead choose real internet access for a single run. The sample then really connects to whatever it contacts — which for malware is often infrastructure its operator runs — from our address, and its DNS lookups go to DNS4EU's unfiltered public resolver (86.54.11.100), run in the EU by Whalebone, Brno, Czech Republic. The run records which mode it used.

9. Verdicts and scoring

The platform assigns verdicts to files and reputation scores to indicators. These are assessments of technical artefacts — a binary, a domain, an address — not of people, and they are computed only from evidence this installation already holds.

Every score is broken down into the named signals that produced it, each shown with the evidence behind it, so an analyst can disagree with the number. An indicator about which nothing is known is reported as unknown, never as benign.

No automated decision-making producing legal effects on a person, or similarly significantly affecting a person, within the meaning of Art. 22 GDPR, takes place. The platform's output informs an analyst; it does not decide anything about anyone.

10. Audit log

Security-relevant actions — signing in, uploading, submitting a scan, inviting or removing a user, changing rules — are recorded with the username that performed them, the action, what was acted on, and the time. Administrators in your organisation can read this log.

This is a security control, and it is also what makes the platform accountable: without it, nobody could establish who did what. It records usernames and actions; it does not record your IP address.

Legal basis: Art. 6(1)(f) GDPR — the legitimate interest of your organisation and of ourselves in the integrity, traceability and security of a system that holds malware and incident data. Art. 32(1) GDPR requires appropriate security measures, of which this is one.

11. Technical and server data

Rate limiting. To defend the sign-in endpoint against password guessing, the platform counts recent requests per IP address in an in-memory store. The counter is keyed by IP address and expires automatically after the length of the counting window, currently 60 seconds. It is not written to the database and not used for any other purpose.

Server logs. Requests are logged in two places, for the operation, debugging and security of the service. The server that terminates HTTPS (Section 13) records each request — timestamp, requested path, status code, your IP address and user agent — and keeps these logs for 7 days before deleting them: long enough to investigate an incident that has just occurred, short enough that no long-term record of individual access is built up. The platform's own services also log the requests they handle — timestamp, path and status code, but not your IP address, which never reaches them. Those logs are rotated by size rather than by date, so they hold only the most recent activity and are overwritten as new requests arrive.

Legal basis: Art. 6(1)(f) GDPR — our legitimate interest in a secure, functioning service. Art. 32 GDPR obliges us to protect it.

12. What leaves the platform

This is a closed list, and the platform is designed to keep it short.

Your samples are never sent to a third-party scanning service. Uploaded files and their contents do not leave this installation. There is no VirusTotal submission, no cloud sandbox, no third-party reputation lookup of the indicators you investigate.

Outbound connections are limited to these:

To whomWhat is sent
The website you submit to the URL sandbox An ordinary page request from our infrastructure — see Section 8
Domain and IP registries, via RDAP (the structured successor to WHOIS), where enabled The domain or address being looked up. This tells the registry that someone is asking about that domain. Queries go to the authoritative registry directly, never through an aggregator, so no additional party sees them; results are cached so the same indicator is not queried repeatedly. It is on on this installation (RDAP_ENABLED) and can be switched off. A registry is a neutral party: it learns that somebody asked about a domain, not who, and not why
The authoritative nameserver of a domain you look up, where active DNS resolution is enabled The domain being resolved. This is the most revealing query the platform makes, and the one to read carefully: for a command-and-control domain the authoritative nameserver is infrastructure the adversary operates, so resolving it tells them that domain is being examined. It is on by default (ACTIVE_DNS_ENABLED), because an indicator you cannot resolve is one an analyst will look up somewhere else — somewhere that also records who asked. Results are cached, so opening the same indicator repeatedly is not repeatedly audible. Like every DNS lookup, the query travels through the recursive resolver our servers are configured with, which therefore sees the domain name too. Switch it off if your threat model includes tipping off the operator of a domain you are investigating; the platform then uses only passive DNS — addresses observed because traffic actually went there during your own scans, which costs nothing extra to learn
Public threat-feed publishers — abuse.ch (URLhaus, ThreatFox, Feodo Tracker) and the Tor Project Nothing about you or your data. The platform downloads the same public files every other subscriber receives, on a nightly schedule that has no relationship to anything you submitted. No indicator, hash or filename is sent in return. Enabled by default; can be switched off (THREAT_FEEDS_ENABLED)
Curated detection-rule publishers — YARA-Forge, via GitHub releases Nothing about you or your data. The platform downloads the same rule package every other subscriber receives, on a weekly schedule with no relationship to anything you submitted. No sample, hash, filename or rule name is sent in return, and no query is made when a rule matches. Enabled by default; can be switched off (RULE_FEEDS_ENABLED)
Antivirus signature distributors (ClamAV and SaneSecurity) and the GeoIP database provider (DB-IP) Nothing about you or your data. The platform downloads signature updates; no sample, hash or filename is ever uploaded in return
The configured outbound mail server, for invitations and password resets The recipient's email address and the invitation or reset message (Section 5)
The configured outbound mail server, for the watchlist digest, to an address your own administrator nominated The values your organisation asked to be alerted about and where each was seen — which may include a sender address taken from a message reported to you. Nothing is sent unless an administrator sets that address; it is empty by default, and clearing it stops the mail. No sample, no file content and no case note is ever included
Nobody, for regulatory reports. NIS2 and DORA drafts are prepared and downloaded by you Nothing. Threatera does not transmit to any supervisory authority, CSIRT or regulator. It renders a document from your own case and tracks the deadline; filing is yours to do, through the channel the authority operates
The TAXII 2.1 or MISP servers you subscribe to, where you have configured any A request for everything the publisher has added since your last poll, and the credential you configured. No indicator, hash or filename is sent — the request names nothing you hold, so the publisher learns only that you are subscribed and when you last collected, which they already know. Off unless configured (Section 12)
The mailbox you nominate for reported phishing, where that feature is enabled The credentials you configured, to read your own mail. Nothing about your data is disclosed: this is an inbound fetch from a mailbox you already control, not a question asked of a third party. Off unless configured (EMAIL_INGEST_ENABLED, Section 7)
The webhook or syslog destination your administrator configures, to forward findings into your own SIEM or ticketing Each finding as it happens — a watchlist hit, a hunt match, a new or changed case, a finished sandbox run, a file judged malicious — with what it is about: the indicator value, the file's hash and the name it was uploaded under, the verdict, the rule or watched value that matched. A webhook is signed with a secret you set; syslog is a CEF or JSON line. No sample and no file content. Sent only to the destination you configured, which must be an address outside our infrastructure; nothing is forwarded until an administrator configures one, and the installation can turn forwarding off altogether (EVENT_SINKS_ENABLED)
Your organisation's endpoint-protection platform (Microsoft Defender for Endpoint, CrowdStrike Falcon or Palo Alto Cortex XDR), where connected A request for the detections raised since the last poll, and for the files those detections name, with a credential your organisation issued, against your own tenancy — your data returning to you. The requests name only what your endpoint platform itself reported; nothing you hold in Threatera is sent to it. Connected only by us at your organisation's request, and off on this installation (EDR_ENABLED)
The identity provider your organisation configures for single sign-on, where configured The sign-in itself: the platform reads the provider's published configuration and signing keys, and exchanges the one-time code your browser brings back for the identity token, using the client credentials your administrator configured. Nothing about your data is sent. Used only for email domains a platform operator has verified as your organisation's, and only when a provider is configured
A key-management service your own organisation runs, where you have configured customer-held keys One 32-byte wrapped key, and nothing else — no indicator, no hash, no filename, no file content. The request asks your own KMS to unwrap your own key so that your samples can be read; the party answering it is you. Off unless configured (BYOK_ENABLED), and the request body is asserted against this description by a test rather than by this paragraph
Whatever a detonated file contacts, when an analyst runs it with real internet access Whatever the sample sends, from our address — the platform does not shape it — and the sample's DNS lookups, through DNS4EU's unfiltered public resolver (86.54.11.100), run in the EU. Off unless chosen for a single run; the default sinkhole sends nothing (Section 8)
netcup GmbH, as processor, which operates the server that terminates HTTPS Every request and response passes through that server, and is decrypted there while it is passed on — see Section 13. It keeps the request logs described in Section 11 and stores none of your content

Geolocation and network-operator enrichment, capability analysis, YARA and Sigma matching — including rules imported from the packages above — passive DNS, threat-feed matching and reputation scoring all run locally, against databases stored on our own infrastructure and against your own observations. Opening an indicator involves no other outbound query: the two named above, RDAP and active DNS, run when you open one, unless a recent answer is already cached.

The distinction that matters here is direction. Downloading a signature set, a GeoIP database or a public blocklist reveals nothing — it is the same file everyone else receives, and it is fetched on a schedule, not when you look something up. Reading your own mailbox reveals nothing to anyone but you. Asking a third party about an indicator is different, because the question itself discloses what you are investigating. Only two features in this platform do the latter — RDAP and active DNS resolution — and on this installation both are on. Two more contact the thing you are investigating directly, and only when you ask: the URL sandbox, and a detonation run with real internet access. They are named here rather than left to be inferred, because this is the fact in this section a reader most needs before deciding whether our configuration suits them.

Of the two, active DNS is the one to weigh. RDAP asks a domain's registry, which is a neutral party and learns only that somebody is curious. Active DNS asks the domain's own authoritative nameserver, and for a command-and-control domain that server belongs to the adversary — so the lookup tells them you are looking. Both can be switched off; switching active DNS off costs you only freshness, because the platform falls back to addresses it observed during your own scans.

We do not transfer personal data to third countries outside the EU/EEA. We do not sell personal data, and we do not use customer content to train machine-learning models. Disclosure to a public authority would occur only where we are legally obliged, and we will inform you unless prohibited from doing so.

13. Hosting and location

The platform and all of its data stores — database, object storage, search index and queue — run on hardware the controller owns and operates himself, at his premises in Austria. No cloud provider holds your data.

Connections reach it through one server operated by netcup GmbH, Emmy-Noether-Straße 10, 76131 Karlsruhe, Germany, located in Vienna and acting as our processor under an Art. 28 GDPR agreement. That server terminates the HTTPS connection and forwards each request over an encrypted WireGuard tunnel, so traffic is encrypted on both legs of its journey; on the netcup server itself it is necessarily decrypted while it is passed on. It stores none of your content.

Being European is a design constraint of this product, not an incidental fact about where a server happens to sit. The controller and the hardware are in Austria, the one provider in the path of your requests is an EU company with its server in Austria, and the data is held in the same country as the supervisory authority responsible for it. No part of the service runs on a non-EU cloud provider, so it is not exposed to a foreign jurisdiction's access powers.

14. Retention and deletion

Account data is kept while your account exists. When an administrator deletes your account it is removed; audit entries recording actions already taken are retained, because a log that can be edited by deleting its subject is not a log.

Invitations expire automatically and can be revoked before they are accepted.

Reported email (Section 7) is retained on the same terms as anything else your organisation holds here, which is worth stating separately because it is the one store that fills with data about people who never chose to be in it. If you enable a mailbox, agree a deletion period for it in the data processing agreement. Deleting your organisation removes the messages, their attachments and the mailbox configuration with everything else.

Content submitted for analysis — samples, scans, indicators, results — is retained for as long as your organisation's retention period says, so that historical investigations remain reproducible and so that an indicator seen today can be matched against a sample seen last year. That is the point of a threat intelligence platform, and it is also a genuine tension with the principle of storage limitation in Art. 5(1)(e) GDPR. The period is yours to choose, which is where that tension is resolved.

Stated precisely: your organisation can set a retention period, and until it does, nothing expires. Every period is unset by default, and unset means keep indefinitely — so a platform nobody has configured deletes no submitted content at all. That default is deliberate: silently discarding evidence would be the worse failure of the two.

Where a period is set, an administrator sets it under Organisation → Data retention, and a nightly job applies it. Deletion of specific material outside that schedule is carried out on request, by us, on your instruction. If your own retention policy requires fixed periods, set them there and record them in the data processing agreement.

Rate-limiting counters expire within a minute. Request logs on the HTTPS server are kept for 7 days; the platform's own logs are overwritten as they rotate (Section 11).

15. Your rights

In respect of the data for which we are the controller — your account and the audit record of your actions — you have the right to:

  • access your personal data (Art. 15)
  • rectification of inaccurate data (Art. 16)
  • erasure (Art. 17)
  • restriction of processing (Art. 18)
  • data portability (Art. 20)
  • object to processing based on legitimate interests (Art. 21)
  • withdraw any consent you have given, with effect for the future

To exercise any of these, contact privacy@threatera.eu. Note that the right to erasure is not absolute: audit entries needed to establish the security and integrity of the system may be retained under Art. 17(3).

If your data is inside content someone uploaded or reported — because your details appeared in a document recovered during an incident, or because a message you sent was reported as suspicious (Section 7) — we are the processor, not the controller, and we cannot lawfully act on your request ourselves. Contact the organisation that operates the Threatera instance, which is the controller. If you contact us instead, we will forward your request to them without undue delay, as Art. 28(3)(e) GDPR requires of us.

You also have the right to lodge a complaint with a supervisory authority. In Austria this is the Data Protection Authority (Österreichische Datenschutzbehörde), dsb.gv.at.

16. Security

Measures under Art. 32 GDPR include:

  • encrypted transport — TLS/HTTPS for all access to the platform, and a WireGuard tunnel between the HTTPS server and the platform;
  • passwords stored as Argon2 hashes and never in recoverable form;
  • uploaded samples encrypted at rest under a key per organisation, for everything stored since that encryption was switched on;
  • multi-factor authentication required for every account that signs in with a password;
  • API keys stored as hashes, revocable individually;
  • mailbox credentials encrypted at rest and never displayed again after entry, since IMAP requires the original password back and it therefore cannot be hashed;
  • four roles per organisation — owner, admin, analyst and read-only — enforced by the server rather than by the interface;
  • each organisation's data separated by row-level security in the database itself, so a query that forgets to filter returns nothing rather than another organisation's rows;
  • an audit log of security-relevant actions (Section 10);
  • rate limiting on authentication to bound credential-guessing (Section 11);
  • URL scanning restricted by an outbound request guard, so a submitted URL cannot be used to reach internal infrastructure;
  • sample detonation confined to an isolated environment;
  • refusal to start with insecure default secrets in a production deployment.

No system is perfectly secure. Where a personal data breach occurs we will notify the supervisory authority under Art. 33 GDPR and, where the breach is likely to result in a high risk to individuals, those individuals under Art. 34. Where we act as processor, we will notify the controller without undue delay under Art. 33(2).

If you believe you have found a security vulnerability in Threatera, please report it to security@threatera.eu.

17. Changes

We may update this policy to reflect changes to the platform or to legal requirements. Where a change materially affects how personal data is processed — in particular any addition to the list in Section 12 — we will inform account holders rather than relying on this page alone. The current version, with its date, always applies and is published here.