Why do data breaches take so long to detect?

By Alan Jeffries, Principal Consultant · Published · Covers the news of the week of 7 September 2026

In September 2026 Japan’s Digital Agency disclosed that an attacker who got into a shared government network in May was not spotted until late June, when a maintenance account was seen opening a large number of files. About 246,000 records may have been leaked. The gap between getting in and being noticed is where much of the damage in a breach is done.

In short

Breaches go unnoticed because attackers often log in with real credentials and behave like legitimate users, and because many organisations do not collect, or do not review, the logs that would show unusual activity. Detection is faster when access to sensitive data is logged and monitored, unusual volumes or times trigger alerts, privileged and maintenance accounts are tightly controlled, and someone actually looks at the alerts.

What was reported

The Star, carrying a Japan News report, said on 11 September 2026 that the attack on the Government Solution Service, a shared government network, started in May after the attacker bypassed a VPN. On 25 June the Digital Agency noticed a maintenance staff account had accessed a large number of files on a server, and on 9 July it confirmed unauthorised access. About 246,000 records may have leaked, covering government staff and businesses or individuals, including names, emails, phone numbers and addresses. The report said the network was built on a zero-trust model but the agency could not prevent the breach.

Why attackers stay hidden

  • They use real accounts, often stolen or misconfigured, so their logins look normal.
  • Maintenance and service accounts often have broad access and little scrutiny.
  • Many systems log logins but not what data was opened or copied.
  • Alerts that do exist are lost among thousands of routine events.

In this case, as reported, it was unusual file access by a maintenance account that first drew attention. A pattern of data access can be an early sign; many breaches, though, are first found by outsiders such as customers, banks or the attacker.

What shortens the gap

  • Log access to sensitive data, not only sign-ins, and keep the logs for months.
  • Set alerts for unusual volumes, unusual hours and accounts touching data they normally do not.
  • Give maintenance and service accounts only the access they need, and review it.
  • Have someone own the alerts, and test the process.
  • Commission a periodic compromise assessment: a forensic look for signs that someone is already inside.

When you find it

The investigation that follows depends on the same logs. See what a forensic investigation establishes and why access logs decide who was affected.

Sources

The Star (Japan News), 11 Sep 2026: Japan government system hit by cyberattack

This article is general commentary on public reports, not legal advice and not a finding about any organisation named in those reports. Facts are as reported at the time of writing.

Frequently asked questions

Why do data breaches take so long to detect?

Attackers often use real credentials and behave like ordinary users, and many organisations do not log or review access to sensitive data, so unusual activity goes unnoticed for weeks or months.

What can be an early sign of a breach?

One sign is unusual activity by a legitimate-looking account, such as large numbers of files opened or data downloaded at odd times. Many breaches, though, are first found by outsiders, such as customers, banks or the attacker.

What is a compromise assessment?

A forensic review of systems and logs looking for signs that an attacker is already present or has been, even when no incident has been reported.

Does zero-trust architecture stop breaches?

It reduces risk but does not eliminate it. Stolen or misused credentials and weak points such as VPNs can still give an attacker access, so monitoring remains essential.

How long should we keep logs?

Long enough to cover the time it may take to detect a breach, which is often months. Take advice on your sector’s requirements and on protecting logs from deletion.

Talk to us before anything is changed.

A short, confidential call is usually enough to tell you what to preserve, what it will cost and whether a forensic examination is worthwhile. We take instructions from solicitors, companies and individuals in Hong Kong and across the Asia-Pacific region.

Telephone +852 5808 1071 · Signal or WhatsApp +852 9103 1852 · Email enquiries@d3forensics.com · Enquiry form