decrypted · 18 november 2026 · ransomware and cybercrime · vulnerabilities and patching · uk policy and law
The patch that broke the network: a week of outages and what officials say
This morning's post covered Java4Shell, a flaw that needs patching. This evening's story is about what happens when the patching itself goes wrong. Since this morning, US officials have given their account of a run of outages that hit Microsoft, air traffic control, banks, Apple and the National Security Agency's internal network in a single week. The account is unglamorous: no coordinated attack, just change going badly. For UK organisations, the lesson concerns how you change systems, not who is attacking them.
A week of failures, and what officials say
Microsoft 365 went down on 12 November, hitting Outlook and Teams in more than 70 countries and territories, with service restored by 14 November. Microsoft traced it to "technical problems associated with an infrastructure upgrade following a network change" and found no sign of malicious activity. Then came a nationwide US air traffic ground stop on 13 November, a software bug slowing Chase bank transactions on 16 November, and an Apple outage on 17 November that the company blamed on an internal update.
Jay Clayton, the US Director of National Intelligence, said on Tuesday: "We do not see the indicia of a coordinated cyber attack." On Wednesday, Secretary of State Marco Rubio said he had "zero evidence" that any of these isolated incidents involved hackers, criminals or another state.
The fix that would not come back up
Clayton's most useful remark concerned the NSA, whose internal network failed on Tuesday. He said the patching that followed a PHP vulnerability "has caused some of these systems to not come back up when you patch them as expected." One outlet called that outage scheduled maintenance, so treat the cause as one official's account.
The PHP flaws are real. The NCSC, the National Cyber Security Centre and part of GCHQ, said on 21 October that two critical issues, CVE-2026-109618 and CVE-2026-109619, may allow attackers to execute commands on any web server running PHP, and urged organisations to find every affected system, internet-facing ones first.
Think of replacing the points on a railway line while trains are still running. The work is not optional, but if nobody has rehearsed how the line reopens, the repair becomes the outage.
Where Secure by Design comes in
Clayton put the structural problem plainly: "Some systems have to hand off to other systems, and those handoffs are not coming as expected." That is a design property, not bad luck. When one system quietly depends on another restarting in the right order, every patch is a gamble on a dependency nobody wrote down.
The design decisions that help are dull and cheap compared with an outage:
- Rehearse patches on a copy of the production system before the real one.
- Keep a tested way back, so a failed update can be reversed in minutes.
- Keep an inventory, because the NCSC also told organisations to check products that bundle PHP, and you cannot patch what you cannot list.
Ask your IT lead when you last tested a rollback for real.
Also this week
Gootloader and the hijacked WordPress site. An FBI flash alert dated 16 November describes Gootloader, malware that uses poisoned search results and compromised websites to gain initial access, and which attackers used as a ransomware precursor. As of October 2026, the FBI said, it had identified over 1,000 infected WordPress sites and 216 domains run by the Gootloader administrator, which led to over $28 million in ransomware payments. The searches it describes targeted legal professionals, with a recent US focus. The alert does not say which ransomware families were involved or what period the figure covers. The UK angle cuts two ways: staff searching for contract wording can be lured, and any organisation with a neglected WordPress site can be quietly turned into the bait. The FBI says the administrator used stolen credentials to reach vulnerable plugins and themes, so plugin updates and administrator credentials are the unfashionable defence.
Ofcom tightens the rules on child abuse material. On 16 November Ofcom set out new Online Safety Act duties. Platforms must ban users who share child sexual abuse material, and hash matching, which compares files against lists of known illegal images, extends to more sites, including pornography services. File-sharing services and the largest social networks must also use automated tools to find previously unknown material. Ofcom expects the duties from early 2027, subject to Parliament. Services under the children's codes must either run highly effective age checks or apply grooming protections to every user: collect more data about everyone, or treat everyone as a child, and neither is free. Ofcom's Oliver Griffiths said "we are not satisfied that enough is being done." The design question for UK providers is how to prove age without building a store of identity documents that becomes tomorrow's breach.
Sources
- Fox Business: Jay Clayton addresses concerns amid widespread tech outages
- IBTimes UK: Rubio says US is 'not under systemic attack' amid outages
- Bitdefender: Microsoft outage forces Japanese workers back to fax
- NCSC: Critical security issue affecting PHP
- FBI FLASH CC-111626-MW: Gootloader Malware Using SEO Poisoning and Compromised Websites
- Ofcom: Platforms must take stronger action to combat child sexual abuse material online
If you want help making your patching and recovery plans boring in the best sense, get in touch.
More like this
- The Lords beat the government on the Cyber Security Bill, and the gap is in design 3 november 2026
- The ICO's verdict on a twelve-year-old test account called Test2 20 october 2026
- The Legal Aid Agency breach reaches back to 1978 16 october 2026
Get the next post by email: subscribe to Decrypted. Double opt-in, unsubscribe any time, or take the RSS feed.
Prefer to listen? Decrypted on Apple Podcasts, or paste the podcast feed into any app.