Legal framework Reference politique-de-securite
Security policy Dated, versioned, public.
Section 01
Our approach
never hold your funds; collect only what is needed, and encrypt what is sensitive; give each component and each person only the access they need; log access and sensitive actions.
Section 02
Your funds and your broker keys
Your funds stay with your broker, in your account, in your name. Aplomb does not hold them and they never pass through Aplomb. In signal mode, we only ask you for a read-only key: you place your own orders with your broker. Your key is encrypted in your browser before it is sent. It is only decrypted when your broker is called, by the one service that calls it, and each decryption is logged. Neither our team nor our administration console is allowed to read it. It is never shown again: the app only shows its last four characters. When it is saved, we check the key's permissions with your broker, when the broker lets us do so: a key that allows a withdrawal, a transfer or trading operations is refused. Its permissions are checked again every day, when the broker allows it; if a withdrawal, transfer or trading permission appears on a read-only key, the key is destroyed, you are notified, and the link with that broker stops until a new key is saved. Some brokers do not let us check a key's permissions. For them, you are the one who declares that the key allows no withdrawal: we cannot check it, and the app warns you about this before you declare it. These keys are only used to read your account. The app shows you the network address our calls come from, so that you can restrict the key to that address at your broker, when it allows it. You can revoke a key at any time, from the app or directly at your broker. If you recover access to your account after losing your second factor, we revoke the keys of your brokerage accounts as a precaution: you connect them again once you are signed in.
Section 03
Protecting your account
Password: it is hashed with Argon2id, an algorithm designed for passwords, and we cannot read it back. When you choose it, we check that it is not among the passwords made public in known breaches, without ever sending it: only the first five characters of its hash are sent to the Pwned Passwords service. Second factor: a passkey or an authenticator app, with backup codes. Text messages are never a second factor at Aplomb. The second factor is required before you connect a broker or have your identity verified. After five wrong codes in a row, the second factor is locked, for longer each time it is locked again, and we warn you from the second lock. Sensitive actions: downloading your data, closing your account, changing your email address, password or phone, and other similar actions require a sign-in less than 5 minutes old, with the strongest factor you have. Session: your sign-in tokens stay on our servers; your browser only receives an opaque identifier. The session closes after 30 minutes of inactivity and at the latest 8 hours after sign-in. You can close all your sessions in a single action. Trusted device: if you choose it, the second factor is no longer asked on that device for 30 days; the password always is. Alerts: we warn you of a sign-in from an unknown device. A change of email address is reported to the old address, which receives a code to cancel it and, for 72 hours, our security alerts. A change of phone number only takes effect after 72 hours. Access recovery: if you lose your second factor, recovery only takes effect after a 72-hour period, during which you are warned and can cancel it. Human verification: at sign-up, password reset, access recovery and after two failed sign-ins within 15 minutes, a Cloudflare check (Turnstile) verifies that the request comes from a person.
Section 04
Protecting your data
In transit: the connection between your browser and Cloudflare is encrypted, and traffic then goes through an encrypted tunnel to our servers. Encryption is enforced on every visit, and our services talk to each other over encrypted connections. At rest: your sensitive information is encrypted by the app under a key specific to your account. Master keys are kept in Oracle Cloud Infrastructure's key vault. Destroying an account's key makes its encrypted data unreadable in our systems; a copy of that key may remain in our backups until they expire, at the latest after about three months, and the erasure register is replayed before any restoration. Identity documents and portraits: you upload them directly to our storage, without going through Cloudflare or our application servers. They are encrypted document by document and are never reachable through a link; only the people on our team in charge of verification view them, on screen, through our console, which is itself routed by Cloudflare, and each viewing is logged. Files: their type and size are limited, and images are re-encoded. The documents you upload, such as an identity document, a portrait or an attachment, cannot be read by anyone before a malware scan has found them clean; a profile photo is checked, then re-encoded. Hosting: your information is stored in Canada, in Toronto, and so are our backups. Monitoring tools: our technical logs and error reports are filtered before sending, to remove personal information. The language model we use to analyse markets only receives market data, never personal information.
Section 05
Our infrastructure
No port of our servers is open to the Internet: traffic arrives through a tunnel that our servers open themselves towards Cloudflare, which also filters attacks. Each component can only reach the services it needs; everything else is refused by default, including outbound Internet traffic. In production, only software we have built and signed, and third-party components whose fingerprint we have checked, are admitted. Before each release, we look for known vulnerabilities in our dependencies: a high or critical vulnerability blocks the release, unless a reasoned exception is recorded. Our secrets (service passwords, keys) are kept in the key vault, never in our code. Our pages apply a strict content security policy, which forbids scripts from elsewhere.
Section 06
Our team's access
Our administration console can only be reached behind a Cloudflare access control, and every action that changes something in it requires a phishing-resistant security key. Each role only opens what it requires. Your information is masked by default in the console; showing it in clear requires a written reason, is logged and is limited in number. Each viewing of your file is logged, and each action requires a written reason. To understand a problem you report, a team member can display your app as you see it, read-only, for a limited time and under a written, logged reason. They cannot change anything in it. At launch, our team has a single person. Actions that would weaken an account's security, such as lifting a suspension, removing a second factor or a passkey, or restoring an email address, only take effect after a 24-hour period, during which the person concerned is warned and can object.
Section 07
Backups and continuity
The database is backed up continuously: in the event of a failure, our objective is to lose no more than 10 minutes of data. Encrypted copies are made every day, and an encrypted copy is kept every week in Canada, outside our host. We plan a restoration test of our backups every month. Before any restoration, we replay the erasure register, so that no erased data reappears. The "Service status" page of the public website shows the state of our functions during a disruption.
Section 08
Security incidents
We record each incident in a register, we assess its risk for the people concerned and we take measures to reduce it. If an incident involving personal information presents a risk of serious injury, within the meaning of Quebec law, or a real risk of significant harm, within the meaning of federal law, we promptly notify Quebec's Commission d'accès à l'information or the Office of the Privacy Commissioner of Canada, as applicable, and the people concerned. In other countries, we notify the authorities their law designates. Our notice says what happened, the information involved, what we did and what you can do to protect yourself. You can report an incident to us in the app, with "Report a security incident". The "Security incidents" section shows the incidents you reported to us and the date of the notice you were given. During an incident, we may suspend a function to protect accounts.
Section 09
Reporting a vulnerability
to use only your own account; never to access, change or delete another person's information; if you access it by accident, stop and report it to us; not to disrupt the service: no flooding attacks, no mass sending, no manipulation of our team or our clients; not to disclose the flaw before we have had a reasonable time to fix it.
Section 10
Protecting yourself against fraud
Our team never calls you, and never contacts you through WhatsApp or Telegram. It never asks for your password, a code or a key. Our support answers you in the app. Each request carries a six-digit verification code: our team quotes it to you and never asks you for yours. Our email notices come from the address [email protected], and their links lead to app.arnoldkom.org. If you have any doubt about a message, do not click: open the app by typing its address yourself, and report the message to [email protected].
Section 11
What we ask of you, and our limits
Choose a password you use nowhere else, and prefer a passkey as your second factor. Keep your devices and browser up to date, and do not tick "Trust this device" on a shared computer. Never share your codes or keys, even with someone who claims to be on our team. At your broker, give your key read access only, and revoke the keys you no longer use. The security of your account at your broker also depends on your broker and on you: follow its own recommendations.
Section 12
Changes to this policy
Versions
Version 1.0
Effective date: October 30, 2026
First version: how we protect your funds and broker keys, your account and your data, our team's access, our backups, what we do in the event of an incident and how to report a vulnerability to us.
Each new version will be dated and summarized here, and earlier versions will stay public on this page.