Written by: Mio Kanzaki(Security Analyst (AI))Reviewed by: 千葉圭介Updated
Phishing prevention means being ready for emails, text messages, phone calls and fake websites that pose as a real company and get you to type in an ID, a password or a card number yourself. The aim is to make such messages less likely to arrive and less harmful when someone is fooled. There are two places to guard: the moment an employee is deceived, and the moment your company's name is faked and your customers are deceived. Three jobs come first: two-factor authentication, email spoofing protection (SPF, DKIM and DMARC), and a rule to confirm any request about money or authentication through a second channel.
This page is a hub for the kinds of phishing, the public figures, the order of settings and what to hand over. If you have already typed something in, read the emergency guide first.
What does phishing actually do to you?
Phishing is less a theft of your key than a trick that makes you hand it over. The National Police Agency describes it as posing as a real organization, leading people from an email or text link to a fake website, and obtaining account details or card numbers.
Fake sites are built to resemble the genuine ones, and you cannot always tell by looking at the screen. So do not rely only on a trained eye. Layer measures that keep damage unlikely even if someone is fooled.
Stolen data is used for unauthorized internet banking transfers, unauthorized card use and unauthorized logins to a workplace system. For a company, the harm comes in two forms: intrusion through an employee's ID, and lost trust when emails carrying your name are faked, which can happen even when you did nothing wrong.
What kinds of phishing are there, and where do you act?
The word phishing covers entrances of different kinds. The table sorts the main methods in the police and Council documents by where you can act. Attackers' wording and steps stay out of this site.
| Method | What happens | Where to act | Read more |
|---|---|---|---|
| Email phishing | An email about an order, refund, identity check or expiring points leads to a fake login page | Staff habits, two-factor authentication, passkeys | Phishing |
| Text message phishing (smishing) | A text posing as a delivery or utility notice leads to a fake site or a malicious app | Opening things from the official app, device settings | Phishing |
| Voice phishing that starts with a phone call | A caller posing as a bank gets your email address, and an email with a link follows | A rule to call back, operating from the official site or app | Phishing |
| Real-time phishing | A code typed into a fake site is entered on the real site at once, defeating two-factor authentication | Authentication that does not work on a fake site, such as passkeys | Missing multi-factor authentication |
| Spoofed email from your own domain | Customers and partners receive fake email in your name | SPF, DKIM and DMARC | Missing email authentication |
| Misuse of your own site's features or settings | A redirect function or a forgotten subdomain becomes an entrance to a fake site | Checking site settings, assessment | Open redirect, subdomain takeover |
The last row is easy to miss: your site's settings can serve an attacker as a stepping stone.
What do public statistics show?
Each number below comes from an original publication that we opened and checked.
National Police Agency: more than 2.45 million reports in a year, with damage flowing to bank accounts
According to the Agency's Reiwa 7 situation report (March 2026), phishing reports in 2025 numbered 2,454,297. The count is the Council of Anti-Phishing Japan's tally, and the report says it keeps rising.
Unauthorized internet banking transfers in 2025 numbered 4,747 cases, with total damage of about JPY 10.397 billion, and phishing accounted for about 90% of the methods. Unauthorized use of credit cards came to about JPY 51 billion (down about 9% from the year before). The White Paper on Police says most of that damage is thought to come from phishing that poses as card companies.
Securities accounts were hit too. From March to May 2025, phishing emails posing as securities firms, unauthorized access to accounts and unauthorized trades surged. The year saw 17,668 unauthorized accesses, 9,824 unauthorized trades and about JPY 740.8 billion in unauthorized buying and selling.
The Agency also notes that since around 2019, real-time phishing that defeats two-factor authentication has been rampant, so a one-time code alone does not stop every method.
National Police Agency: the phone call that starts an attack on corporate accounts
From autumn 2024 to April 2025, unauthorized transfers from corporate accounts through voice phishing (vishing) surged. Criminal groups phone a company, pretend to handle an update to online banking, obtain an email address and send a phishing email. There was no damage from May to October, then a new surge in November that hit many regional small and mid-sized financial institutions.
The clues are an international caller number, an automated voice that switches to a human, and a request for an email address during the call followed by an email with a link. The Agency asks companies to make two things routine: when a bank calls, check by calling back the branch or main number, and open online banking only from the official site or app.
Council of Anti-Phishing Japan: high counts that rise and fall, and a skewed set of brands
In the Council's monthly reports, January 2026 had 202,350 reports and September 2026 had 94,661, which was up about 15.0% on the previous month. September involved 49,032 phishing site URLs (duplicates removed) and 105 abused brands. In that month, messages posing as Amazon were about 33.2% of the total and the top five brands about 65.3%. By sector, e-commerce was about 47.9%.
The Council also reports changes in method: emails that still contained instruction text apparently left by an AI, emails that put the recipient's name in the body, and URLs that abuse domains of legitimate services such as amazonaws.com. It also confirmed support scams that fake the recipient's own domain, and emails aimed at accounting staff posing as a document check.
What DMARC figures show about what works and what does not
In the Council's September 2026 survey, about 37.8% of the phishing emails that reached its survey addresses faked a real service's domain. Of those, about 19.7% came from domains whose DMARC setting asks for reject or quarantine, about 4.3% from domains with a monitor-only "none" policy, and about 13.8% from domains without DMARC.
The other 62.2% or so were sent from domains the attackers registered themselves, and about 98.9% of those passed DMARC authentication. So DMARC addresses the harm of having your name faked, and it cannot stop email that an attacker sends from its own domain. The Council asks businesses to plan a move of the DMARC policy to reject.
A joint announcement by the Ministry of Internal Affairs and Communications, the National Police Agency and the Ministry of Economy, Trade and Industry on 12 March 2026 points the same way. It asks access managers to adopt phishing-resistant authentication such as passkeys, and to introduce DMARC, SPF and DKIM in a planned way, with a DMARC policy that requests rejection of spoofed email.
IPA: it appears on both the individual and organizational lists
In IPA's Top 10 Information Security Threats 2026 (29 January 2026), "theft of personal information through phishing" is one of the threats listed for individuals. For organizations, business email compromise is tenth.
After the figures, you may want to see what your domain's email authentication looks like. The free URL check reads SPF and DMARC records from public information only. It costs JPY 0.
What do SPF, DKIM and DMARC do?
Three settings work together against email spoofing (email domain authentication), each with its own role.
SPF is a list of the servers allowed to send your company's email. DKIM is a digital seal attached to each message, so the receiver can confirm that nobody changed it on the way. DMARC lets the sender tell receivers what to do with email that fails the list or the seal: let it through, quarantine it or reject it.
Do not set DMARC to reject at the start. If you have not found every service that sends email for you, genuine messages will stop arriving. Proceed in this order:
- Write down every service that sends email from your domain: internal mail, newsletters, booking and order notices, and contact forms.
- For each service, register it in SPF and turn on DKIM signing.
- Publish DMARC in monitor-only mode (p=none) and decide where the aggregate reports go.
- Read several weeks of reports, confirm that genuine email passes, and fix any sender that does not.
- Raise the policy to quarantine, then to reject.
Step 3 means adding one TXT record to the domain's DNS. The shape is shown below, with a fictional domain and address.
_dmarc.example.co.jp. TXT "v=DMARC1; p=none; rua=mailto:dmarc-report@example.co.jp"
A DNS mistake can stop your site and email, so note your current settings before you change anything. If you use many outside sending services or own several domains, ask a specialist to look before step 5.
What should you do first, by company size?
Begin with the items that bring the most for your type of company.
A company of a few people up to about 20, renting email as a service
First, two-factor authentication for everyone's email and cloud services, so that a stolen password is not enough. If you can, choose passkeys, which do not work on a fake site. Next, make a rule not to open money or authentication messages from email links. Open notices from banks, couriers and cloud services from a bookmark or the official app. For a request to change a payee or reset a password, confirm the sender by another means, such as a phone call, before acting.
A company of a few dozen, with accounting, administration and sales apart
The Council has confirmed emails aimed at accounting staff that pose as a document check, so accounting and administration can take a short drill that shows them realistic messages. In parallel, publish DMARC in monitor mode and raise it to quarantine and reject as you read the reports. Choose one place to report suspicious email, and run it so that nobody is blamed for opening something. Quick reporting lets you start a response sooner.
A company that sends email in its own name to many customers (shops, memberships, bookings)
Your customers suffer when your name is faked. Set reject as the goal for DMARC, and set up SPF and DKIM for each service that sends your email. Also watch for newly registered domains that resemble yours, and decide whom to contact when you find a fake site, such as the hosting company and the Council of Anti-Phishing Japan. Offering passkeys for your customers' logins also makes damage easier to reduce.
10 things to do today
These 10 steps come from our fix menu and the police and Council documents, arranged so that you can start easily. Fix-menu numbers are given where an item matches one.
- Turn on two-factor authentication for email, cloud services and admin screens. Fix-menu number 3. Prefer passkeys or an authenticator app.
- Tidy administrator accounts and passwords. Fix-menu number 12. Delete accounts nobody needs and end password reuse.
- Check SPF for your domain. Register every service that sends email from it.
- Turn on DKIM signing. Each email delivery service has a signing option in its settings.
- Publish DMARC in monitor-only mode. Decide where the aggregate reports go, and only watch them for a few weeks.
- Open bank, courier and cloud notices from the official app or a bookmark, never from the link. Put it on one sheet for everyone.
- If a caller claims to be your bank, hang up and call back the official main number. The National Police Agency gives this as a measure.
- Confirm payee changes and reset requests by another channel before acting. Stress this with accounting and administration.
- Choose one place to report suspicious email. Agree also that nobody is blamed for reporting a click.
- Inspect unused subdomains, DNS records and your site's redirect functions. Forgotten settings can become an entrance for a fake site.
Begin with items 1 and 2, then move on to 3 to 5. Items 3 to 5 touch DNS, which can stop your site and email if you get it wrong, so write down the current settings first. Doing all 10 does not guarantee that nothing will happen.
Short of time to do it yourself? Ask us and we will mark which of the 10 are still open.
What should you hand to a specialist?
Phishing targets human judgment, so settings alone do not finish the job, and some settings stop the business if set wrongly. A good test is whether a mistake can be undone.
We recommend a specialist for these:
- Designing email domain authentication and moving in stages. This means finding every sending service and ordering the steps up to reject. A server and network assessment is from JPY 49,800 (excl. tax, up to 5 IPs), and it covers the DNS settings.
- Checking holes in your own site. We check misuse of redirect functions and forgotten subdomains in an assessment. A WordPress/CMS site assessment is from JPY 49,800 (excl. tax, up to 30 pages). Page count, login-only screens and paid plugins move the price.
- Designing staff education and drills. Cost depends on headcount and content, so we discuss it in a free consultation.
- Answers to partners. For the question "Do you protect against email spoofing?", the security advisory plan at JPY 29,800 per month (excl. tax) can help you prepare an answer.
- Response after someone has typed something in. Reviewing login history and finding forwarding rules nobody set up need steps that preserve evidence. Start from the emergency guide.
We confirm and fix issues found in an assessment, check settings, and investigate and recover after damage. A PDF listing every change reaches you before delivery, and you confirm the real screens yourself. Free support covers the month after delivery.
We do not sell a spam filter that stops phishing email before it arrives, and we do not promise protection. Methods keep changing, so an assessment confirms only that the weak points known now are absent.
Which pages cover the individual holes?
Holes tied to phishing have their own pages, with the mechanism, checks and fixes.
People and email
- Phishing
- Missing email authentication (SPF, DKIM, DMARC)
- Missing multi-factor authentication
- Password reuse
Site settings that become an entrance to a fake site
For neighbouring topics, see hacking prevention, unauthorized access and data leaks.
How can the free URL check help you decide?
Pick what to fix only after you know where you stand. The free URL check takes your site's address and uses information that anyone can see from outside: the submitted page, public DNS and the encrypted connection. It reports certificate expiry, security headers, software versions that show on the page, and SPF and DMARC records, and it matches the JavaScript libraries the page loads against known vulnerabilities. Anything behind login, and the server itself, stay out of its reach.
It does not send test emails or fake phishing messages to your staff. A drill is designed separately, with your consent. A concerning result can move on to the free assessment, once we have signed an NDA and received your consent form. Tools and a person make a quick check, and an issue report follows. The formal quotation comes after that free assessment.
Frequently asked questions
If we run a spam filter, are we protected from phishing?
A filter reduces it but cannot stop all of it. The Council of Anti-Phishing Japan has seen emails slip past filters by placing invisible characters between words. Combine a habit of not opening links from messages with two-factor authentication, and do not rely on the filter alone.
If we set up DMARC, will phishing emails stop arriving?
Only emails that fake your own domain are affected. In the Council's September 2026 report, about 62.2% of phishing emails came from domains the attacker registered, and about 98.9% of those passed DMARC authentication. Set DMARC up so that your own name cannot be abused.
What if an employee opened a link and typed in an ID and password?
Change the password of the service they typed it into right away, and change it on any other service where it is reused. Then check the login history and look for forwarding rules nobody set up, and report to the person in charge. Not blaming the person who reports leads to earlier reports.
How much does phishing prevention cost?
The free URL check, which looks only at what is visible from outside, costs JPY 0 and also shows whether SPF and DMARC records exist. A server and network assessment is from JPY 49,800 (excl. tax, up to 5 IPs), and the amount depends on how many IP addresses are involved. Training and drill emails depend on headcount and content, so we discuss them in a free consultation.
Next step
Two things for today: start on items 1 and 2 of the list, and check your domain's email authentication with the free URL check.
If you want help ordering the work or deciding what to outsource, talk to us. If you have already typed in a password, or have acted on a suspicious request for a transfer, begin with the emergency guide.
Written by Mio Kanzaki (AI security analyst). Supervised by 千葉圭介. Updated 19 November 2026.
Frequently asked questions
If we run a spam filter, are we protected from phishing?
A filter reduces it but cannot stop all of it. The Council of Anti-Phishing Japan has seen emails slip past filters by placing invisible characters between words. Combine a habit of not opening links from messages with two-factor authentication, and do not rely on the filter alone.
If we set up DMARC, will phishing emails stop arriving?
Only emails that fake your own domain are affected. In the Council's September 2026 report, about 62.2% of phishing emails came from domains the attacker registered, and about 98.9% of those passed DMARC authentication. Set DMARC up so that your own name cannot be abused.
What if an employee opened a link and typed in an ID and password?
Change the password of the service they typed it into right away, and change it on any other service where it is reused. Then check the login history and look for forwarding rules nobody set up, and report to the person in charge. Not blaming the person who reports leads to earlier reports.
How much does phishing prevention cost?
The free URL check, which looks only at what is visible from outside, costs JPY 0 and also shows whether SPF and DMARC records exist. A server and network assessment is from JPY 49,800 (excl. tax, up to 5 IPs), and the amount depends on how many IP addresses are involved. Training and drill emails depend on headcount and content, so we discuss them in a free consultation.