The attacks of 2026, and the habit that stops each.
We track new breaches and the methods behind them every day. Here is what is being used against firms like yours in 2026, what we read from outside, and the habit that stops each.
Phishing written for the AI that reads the inbox
An inbox summary is a reading of the email, not a check of it.
What it is
An email carries a second message that no person sees: text made invisible, a note in the code, or characters that take up no space. It tells the AI that sorts or summarises the inbox to call the email urgent and genuine, or to say that the payment details have changed.
What we read from outside
On a paid assessment and on monitoring, our deep read of your firm's own website lists text hidden from visitors that gives instructions to AI tools, so your own site does not carry the same trick. What we cover
We never open your mailboxes, so the email itself is read by your team.
What your team does
- Read the email itself before acting on a summary of it whenever money, a password or a file is asked for.
- Confirm any request to pay or to change bank details on a number you already hold, never one in the email.
- Let no AI tool pay, send or forward on its own. A person approves each one.
One person and an AI, at the scale of a hundred hackers
One attacker with AI tools now does the work of a crowd.
What it is
A single attacker now uses AI tools to search the internet for exposed systems and get in within minutes. Once inside, the same tools automate the rest of the attack.
What we read from outside
We read the doors these tools try first: old software with a known, published flaw, open ports and remote access, sign-in and admin pages anyone can reach, private files and keys left public, and your firm's addresses in monitored breach databases. What we cover
How fast we look again, on monitoring: email settings, look-alikes and new host names re-checked every hour, the rest every day.
When software your website shows lands on the US government's list of flaws being used in attacks, a firm on monitoring is told the day our daily match finds it.
What your team does
- Turn automatic updates on for every computer, phone, website and server.What changes: updates install by themselves, now and then with a restart. It stays on. Choose the hour they run in the update settings, and pause one there if it breaks something.
- Leave nothing that runs your systems reachable from the internet: admin pages, remote desktop and file shares sit behind a private connection or are switched off.What changes: staff connect through the private connection before they reach those pages. It stays on. A page can be opened to one address when it is needed, and closed again the same way.
- Turn on a second sign-in step for every account, email first.What changes: a code or a tap on the phone at each new sign-in. It stays on. A backup code or a second device covers a lost phone, and the method is changed in the account's security settings.
- Remove the accounts nobody uses: former staff, old suppliers, test logins.
Stolen logins at a supplier
The email that empties a firm's account often comes from a real supplier's real mailbox.
What it is
An attacker signs in to a supplier's or a client's mailbox with a stolen password and writes from inside it. The thread is real, so the new bank details in it look real.
What we read from outside
We read whether addresses at your firm's own domain are listed in monitored breach databases, and we count each one as an open way in. What we cover
2 in 10 professional firms have a shared address such as info@ in monitored breach databases. Measured 30 Sep 2026. Read the study
On monitoring, you name up to 5 suppliers or clients, and we email you the day a new domain made from one of their names appears and is worth a look. Before money moves
We do not read your suppliers' mailboxes.
What your team does
- Confirm every change of bank details on a number you already hold, even inside a real email thread.
- Give every work account a second sign-in step, and never use a work password anywhere else.What changes: a code or a tap on the phone at each new sign-in. It stays on. A backup code or a second device covers a lost phone, and the method is changed in the account's security settings.
- Agree with your main suppliers how a change of bank details is confirmed, and keep it in writing.
Deepfake voice and video calls asking for payment
A familiar voice is no longer proof of who is speaking.
What it is
A call or a video meeting uses a copied voice or face of a partner, a client or a supplier. It asks for an urgent payment or a change of bank details, often against a short deadline.
What we read from outside
We do not read this from outside.
On monitoring, the call-back sheet lists the number each supplier you name publishes on its own website, with the date we read it. Before paying changed bank details, call this number, never one given in the email. Before money moves
What your team does
- Hang up and call back on a number you already hold before any money moves.
- Agree a word or a question with the people who approve payments, and ask for it on any unusual request.
- Have two people approve every new payee and every change of bank details.
QR-code phishing
A code in a picture slips past the checks that read links.
What it is
An email, a letter or a sticker carries a QR code instead of a link. The phone that opens it lands on a copy of a sign-in page, away from the protections on the work computer.
What we read from outside
We do not read this from outside.
What your team does
- Never sign in to a work account from a QR code. Type the address you know instead.
- Treat a QR code in an email as a link from a stranger.
- Tell the rest of the firm when one arrives, so nobody else opens it.
Sign-in prompt floods
A tired yes is the easiest way past a second sign-in step.
What it is
An attacker who already has a password sends approval request after approval request to a person's phone, often late at night. One tap on approve lets them in.
What we read from outside
We do not read this from outside.
On a paid assessment and on monitoring, our deep read lists the sign-in pages your own website leaves open to anyone, so you know which doors face the internet. What we cover
What your team does
- Never approve a sign-in you did not start. A prompt you did not start means your password is known.
- Change the password of any account that gets prompts nobody started, the same day.
- Use a second sign-in step that asks for a number shown on the screen, not a single approve button.What changes: the person types the number the sign-in screen shows instead of tapping approve. It stays on, and the method is chosen in the account's security settings.
Fake job ads and fake apps in a firm's name
A firm's name is now used to hire people and to publish apps the firm has never seen.
What it is
A job ad in the firm's name sends applicants to a free email address or a messaging app, then asks for identity documents or a payment for equipment. A fake app uses the firm's name and logo to collect clients' details.
What we read from outside
We look in the app stores and on the job sites for an app or an ad that uses your firm's name but is not yours. A person at LeakTrace checks each one before you see it, and we give you the steps to have it removed.
Part of the paid assessment and of monitoring. What you get
What your team does
- Say on your own website where your firm posts jobs and which apps are yours.
- Tell applicants and clients that your firm never asks for identity documents or payment by text or chat.
- Search your firm's name in the app stores and on the job sites now and then.
Look-alike domains
One changed letter is enough to borrow a firm's name.
What it is
Someone registers a domain one letter off your own, or your name on another ending, and sets it up to send email. Clients and staff read the name they expect to see.
What we read from outside
We read the registered look-alikes of your firm's domain and list which can send or receive email or carry a website. New ones looked for every hour on monitoring. Look-alike domains
4 in 10 professional firms have a look-alike of their domain registered and live. Measured 30 Sep 2026. Read the study
What your team does
- Read the whole sender address on any email that asks for money or a password.
- Check the address of a new request against one you already have on file before you reply.
- Register the obvious copies of your own domain before someone else does.
Forged email in your firm's name
The cheapest attack on a firm is an email in its own name.
What it is
Anyone can send an email that shows your firm's address as the sender. Unless your domain publishes a rule telling mail servers to refuse it, the forgery can reach your clients' inboxes.
What we read from outside
We read the DMARC and SPF records of your firm's domain and of every other domain we find your firm owns, because each domain needs its own rule. Re-checked every hour on monitoring. What we cover
3 in 4 of the professional firms we read have no published rule telling mail servers to refuse mail forged in their name. LeakTrace index, readings to 8 October 2026. Read the index
The one change we can make ourselves is to your email records, and only when you approve each step, where your DNS host allows it.
What your team does
- Treat an email from a colleague that asks for a payment or a password as unconfirmed until you have spoken to them.
- Tell your clients how your firm confirms bank details, so a forged email that asks otherwise stands out.
- Keep a list of every domain your firm owns, the old ones too.
Ads aimed at one person
An ad can now be bought for an audience of one.
What it is
Ad networks let a buyer choose who sees an ad. Spyware sellers have used that to send a booby-trapped ad to one chosen person, often an owner, a partner or a family-office principal, and in some cases the ad only had to load, with no click.
What we read from outside
We do not read this from outside.
It happens on the person's own phone and laptop.
What your team does
- Keep phones and laptops updated, with automatic updates on.
- For the people most likely to be targeted, turn on the phone maker's highest protection mode: Apple's Lockdown Mode, and on Android, Google's Advanced Protection. Apple's help page on Lockdown Mode. Google's help page on Advanced Protection.What changes: on an iPhone most message attachments and link previews are blocked and some websites load slowly or break; on Android apps from outside the store cannot be installed and the phone will not use 2G networks. Both are made to stay on. A trusted website or app can be excluded from Lockdown Mode, and each is turned off at the same setting that turned it on.
- Use a content and ad blocker, or a filtering DNS service, on work devices.What changes: ads and known bad sites do not load, and now and then a site you need is blocked too. It can stay on. Allow that site in the blocker's or the service's settings.
- Restart phones regularly.
What we read, and what your team confirms inside.
| The attack | What we read from outside | What your team confirms inside |
|---|---|---|
| Phishing written for the AI that reads the inbox | Text hidden for AI tools on your own website, on a paid assessment and on monitoring. | A person reads the email itself and approves every payment, send and forward. |
| One person and an AI, at the scale of a hundred hackers | Old software, open ports and remote access, reachable admin pages, files left public, breach-listed addresses. | Updates on, admin pages off the internet, a second sign-in step, old accounts removed. |
| Stolen logins at a supplier | Your firm's own addresses in monitored breach databases. | Bank detail changes confirmed on a known number; a second sign-in step on every account. |
| Deepfake voice and video calls asking for payment | Nothing. | Call-backs on a known number, an agreed word, two approvals for a new payee. |
| QR-code phishing | Nothing. | No sign-in from a QR code; the address typed by hand. |
| Sign-in prompt floods | The sign-in pages your website leaves open, on a paid assessment and on monitoring. | No approval of a sign-in nobody started; the password changed the same day. |
| Fake job ads and fake apps in a firm's name | Apps and job ads in your firm's name, checked by a person, on a paid assessment and on monitoring. | Where you post jobs and which apps are yours, said on your own website. |
| Look-alike domains | Registered look-alikes of your domain, and which can send email or carry a website. | The whole sender address read before money or a password moves. |
| Forged email in your firm's name | The rule each domain your firm owns publishes against forged email. | Requests for money or a password confirmed by speaking to the person. |
| Ads aimed at one person | Nothing. | Devices updated; the highest protection mode on for the people most likely to be targeted. |
New breaches, read every day.
See what your firm shows the internet, without handing over a key.
Your domain and a work email at it. No password, no login and no software. Read-only, and sent only to your work email.