The evidence brief · sourced digital forensics news · updated 16 November 2026
Digital forensics news archive
Every story in the brief, newest first, grouped by the month the source was published. Filter by desk, and follow any headline to the report it is based on. 17 stories so far (Courts & evidence: 2 · Breaches & incidents: 6 · Investigations & enforcement: 4 · Threats & tooling: 2 · Policy & regulation: 3).
Cyber Daily reports that Cyber Security NSW is responding to a breach affecting an isolated NSW Health server, declared a major incident and under investigation by the NSW Police Cyber Crime Squad. The report says no patient data has been affected at this time. A hacker on a forum claims to hold "100,000 sensitive documents"; that claim has not been verified.
Our note: Claims on hacking forums about stolen volumes can be inflated or recycled, so they need comparing against what the logs show.
Source: Cyber Daily, 4 November 2026 · added here 16 November 2026
Federal prosecutors in the Northern District of Texas unsealed a 22-count indictment alleging a bank account takeover conspiracy from July 2023 to November 2024, said to have used SIM swapping and corrupt bank and telecoms insiders. Prosecutors say a spreadsheet listed nearly 3,000 alleged victims whose accounts together held over US$373 million; that is a balance, not a stated loss. These are allegations, and the defendants are presumed innocent.
Our note: SIM-swap investigations typically draw on carrier SIM-change and staff-access records alongside bank login data. In the US, carrier records usually need a subpoena or court order.
OPSSHIELD says an attacker entered its billing system through a customer account between 1 and 2 November and pushed a fake cPGuard update through its update channel, live for about 70 minutes. About 2,950 customer servers ran the update and about 7,900 billing accounts were exposed; billing passwords were stored as bcrypt hashes.
Our note: With a malicious vendor update, each customer needs its own logs and host artefacts to show whether the update ran, and analysis of the payload to show what it did.
The National Cyber Security Centre's 2026 annual review sets out seven lessons from 337 incidents, including that attackers exploit the basics, people remain the target and early action reduces impact. The review says: "Most successful attacks continue to exploit familiar weaknesses." The figures cover only incidents the NCSC handled, not every UK intrusion.
Our note: If familiar entry points keep recurring, collecting logs from internet-facing systems and identity services early is a sensible first step in most investigations.
Eurojust and Europol supported action against an online investment fraud network that used legitimate-looking trading platforms, with more than 100 servers shut down and seven arrests. Estimated losses are over EUR 4 million. Those arrested have not been convicted and are presumed innocent.
Our note: Seized servers and platform records, together with bank, payment-provider and crypto-exchange records, help link victims' payments to the people behind a fake trading site.
Source: Eurojust, 3 November 2026 · added here 16 November 2026
The UK government updated its factsheets for the Cyber Security and Resilience (Network and Information Systems) Bill, which was introduced to Parliament on 12 November 2025, and added two new ones on agentic AI resilience measures and vendor risk management. The factsheets explain the measures the Bill proposes.
Our note: If the Bill is enacted as proposed, its incident-reporting deadlines would shape how quickly organisations in scope need to gather evidence about an incident.
Source: GOV.UK, updated 3 November 2026 (first published 12 November 2025) · added here 16 November 2026
Bitdefender describes a 3AM ransomware affiliate breaching a UK school estate and using built-in Windows tools to reach other schools. It calls the attack hybrid, part human-operated and part what looks like an automated agentic loop, but says that mechanism is not proven. A machine list described in the report held 47 systems.
Our note: Attacks that use built-in admin tools leave few malware files, so command-line, authentication and remote-access logs (where they were switched on) and Windows system artefacts carry the timeline.
Source: Bitdefender, 3 November 2026 · added here 16 November 2026
Hong Kong's Privacy Commissioner for Personal Data (PCPD) opened formal investigations into breach notifications from Cathay Pacific, Hong Kong Airlines, Malaysia Airlines Berhad, YOHO Hong Kong and Broadway, The Standard reports. The paper says the PCPD had received dozens of complaints and enquiries, with reported losses from HK$10,000 to more than HK$1.5 million in scams the paper links to the breaches. The investigations have not made any finding.
Our note: When scammers quote real booking or order details, linking a fraud to a particular breach means matching the victim's records against what each leak contained.
If you are affected: If you are a customer of one of these companies, treat any call or message that quotes your booking or order details with suspicion, check it through the company's published contact details, and never share passwords or one-time codes or transfer money on request.
At a press conference on 2 November, GMO Internet Group's chairman said the company had failed to address security flaws flagged two years earlier, according to Jiji Press. The group had been warned in April 2024 by GMO Cybersecurity by Ierae, a group company. The chairman's pay is being cut by 30 per cent for three months.
Our note: Earlier internal security findings often become important evidence when regulators and claimants ask what a company knew and when.
From 2 November 2026, outgoing calls from the Singapore Police Force show one caller ID, 6000 0999, a change the force describes as a step to combat scams. Caller ID can be spoofed, so the number on its own does not prove a call is genuine.
Our note: A spoofed number often appears in the receiving carrier's records too, so tracing a call usually needs records from the originating or interconnecting network, obtained through the proper legal route.
Club Tourism said it confirmed unauthorised access to part of its server on 30 October and that customer details such as names, addresses, phone numbers, emails, dates of birth and emergency contacts may have leaked. It puts the number of records at about 4.32 million and says passwords, passport, card and bank details were not included. It has notified the police and Japan's Personal Information Protection Commission.
Our note: An early 'may have leaked' notice usually comes before a forensic review settles what was actually taken, so the figures may change.
The Divisional Court dismissed challenges by Meta, X and TikTok to information notices Ofcom issued under section 100 of the Online Safety Act 2023 ([2026] EWHC 2779 (Admin)). Blackstone Chambers' summary says four grounds were found unarguable and the proportionality ground failed. The case concerns Ofcom's information-gathering powers, not any finding that the platforms breached their safety duties.
Our note: The ruling concerns Ofcom's power under the Act to require information from platforms; it does not decide what that information will show.
TechCrunch reports that a database credential built into Applebee's Android app, which has more than a million downloads, gave access to customer data including names, emails, dates of birth and phone numbers. The number of people affected is not known, and Dine Brands said it had not found any evidence of unauthorised access.
Our note: As a general point, a finding of 'no evidence of unauthorised access' is only as strong as the logging behind it, so an early question in any such review is how far back the access logs go.
Source: TechCrunch, 30 October 2026 · added here 16 November 2026
Representatives of OpenAI and Meta told the US Judicial Conference's Advisory Committee on Evidence Rules that checking whether audio and video evidence is genuine will become more difficult. The report cites a Federal Judicial Center survey in which only 15 of 931 judges said they had dealt with a deepfake challenge.
Our note: The US committee is weighing how audio and video should be authenticated, so examiners may see more requests to test where a recording came from, not only what it shows.
EDRM reports on an Iowa Supreme Court decision of 9 October 2026 that upheld a conviction after a video was admitted over the defence's objection that it was a deepfake. The court found the claim speculative and accepted circumstantial authentication, even though there was no chain-of-custody evidence linking the video to the defendant.
Our note: In this Iowa case, an unsupported deepfake objection was not enough. A party relying on a recording may still want to be ready to authenticate it from its content, context and metadata.
Ahold Delhaize said it was experiencing a security issue within its network and had taken some systems offline as a precaution, while its stores stayed open. The statement gave no cause and no figures.
Our note: As a general point, isolating systems from the network can preserve volatile memory, which is lost if they are powered down, so memory and log capture usually come before any shutdown.
St Andrew's Hospital in South Australia isolated its systems after a subset of data was downloaded from its servers, MedRisk reports. The hospital has alerted the Australian Cyber Security Centre and the Office of the Australian Information Commissioner (OAIC). No count of affected patients was given.
Our note: Health breaches turn on exactly which records were taken. Server, endpoint and network logs are the usual starting point, though they may not settle it on their own.
Secondary report: MedRisk, 10 October 2026 · added here 16 November 2026