Open through Christmas & New Year — minimum days off. Independent cyber cover while your team is away. Holiday readiness →
← All insights
Cyber Risk & Strategy October 24, 2026 9 min read

PHP2Shell Took Government Websites Offline. Five Questions for Your Board Before Anyone Signs Off on ‘We’ve Patched’

On Wednesday night, cyber.gov.au went dark. So did a number of other Commonwealth websites, after the Australian Signals Directorate’s Australian Cyber Security Centre issued a critical alert for two flaws in PHP, the scripting language behind a large share of the world’s websites. The flaws, now known as PHP2Shell, allow an attacker to run code on a web server without logging in. For a board, the useful question is not whether IT has applied the update. It is whether anyone can prove the organisation knows every place PHP runs, who patched each one, and that nobody got in first.

CISO Advisory Australia gives boards an independent read on exactly that kind of question. If you want someone senior to test the answers you are getting this week, talk to an independent CISO.

What happened, and why governments pulled the plug

The two vulnerabilities are CVE-2026-109618 and CVE-2026-109619. According to the PHP project’s security release, they “can allow unauthenticated remote code execution”, and the fixed versions are PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35. PHP says every installation serving web requests is at risk, including PHP-FPM, Apache with mod_php, php-cgi, FrankenPHP and LiteSpeed, with the highest risk on mod_php.

ASD’s alert says all PHP versions from PHP 5 onwards may be affected. Its advice is short: identify PHP in your environment, follow PHP’s official guidance, apply updates and other appropriate mitigations, and monitor PHP applications and related systems for suspicious activity. ASD also said it was not aware of successful active exploitation within Australia.

That last point needs to sit next to what was being reported elsewhere. The ABC reported that Cloudflare’s chief technology officer said early attacks were emerging and described Cloudflare’s firewall rules as a stopgap only. iTWire reported that Amazon’s threat intelligence team logged an exploitation attempt on 17 October, four days before disclosure. Both things can be true. No confirmed Australian victim does not mean no Australian exposure.

The disclosure itself was messy. iTWire reports that the CVE numbers leaked on 20 October, forcing the release forward a day, and that one vendor published patched container images hours before the coordinated release. Several firms that found the bugs said AI tools did much of the work of turning a long-standing flaw into a working exploit, and the PHP Foundation acknowledged that one of the issues had first been spotted in a 2024 audit.

The most instructive part of this week was not the bug. It was the decision several governments made before the patch existed. The ABC reported that agencies in the Netherlands and Finland took sites offline as a precaution, that France’s Interior Ministry pulled local government administrative websites, and that a Danish municipal IT provider shut down services including a national school communication platform. Information Age found the Australian Bureau of Statistics and Jobs and Skills Australia unavailable on Thursday morning, both back soon after.

A spokesperson for Home Affairs Minister Tony Burke said “some websites are temporarily unavailable as work is undertaken to mitigate risks arising from this vulnerability.”

Taking a public website offline at night is a business decision, not a technical one. It costs revenue, customer trust and sometimes service obligations. Doing it well requires that someone already has the authority to make the call, that the criteria are agreed in advance, and that the board will back the person who makes it. An incident response plan that only describes recovery leaves the harder question open: who may switch off a customer-facing service on suspicion alone, before anything has gone wrong?

That gap is cheap to fix and expensive to discover at 11pm.

Question one: do we actually know where PHP runs?

This sounds like the easy one. It rarely is.

PHP sits under WordPress, Drupal, Joomla and Magento, and iTWire notes it also underpins GovCMS, the Drupal platform behind much of the federal web presence. It also turns up where nobody thinks to look: hosting control panels, vendor appliances, container images, old marketing microsites and the admin portal a supplier built three years ago. Qualys’s threat research lead, quoted by iTWire, advised inventorying hidden PHP in containers, cloud images, hosting panels, secondary drives and vendor appliances.

The Australian Information Security Association put it bluntly in Information Age: “If you don’t know what software you’re running, you can’t know whether you’re vulnerable.”

A good answer from management names the systems, the owner of each, the PHP version before and after, and the date and time each was fixed. A weak answer is “our website is on WordPress and it’s been updated”. Updating WordPress does nothing here, because the flaw is in the PHP runtime underneath it, a point several responders made this week.

Need an independent CISO on this, this week?

CISO Advisory Australia provides senior, independent security leadership to boards and executives, with nothing to sell but advice. Talk to us about a virtual CISO retainer, an Essential Eight assessment or an urgent review.

Talk to an independent CISO → 24/7: 07 2112 8502

Question two: who patches the parts we don’t run ourselves?

A large share of Australian organisations do not operate their own web servers. The website is with an agency, the booking portal is a SaaS product, and the hosting is shared. In each case the PHP runtime belongs to someone else.

That makes this week a contract test. Can you get, in writing, confirmation from each provider that their PHP is on a fixed version, with a timestamp? Does your contract oblige them to tell you about a critical vulnerability, and how fast? A UNSW researcher quoted by iTWire made the supply chain point directly: even if an organisation patches its own website, a third-party component running vulnerable PHP remains exploitable.

If the honest answer is that nobody knows who is responsible for the runtime on a given system, record that as a governance finding. It will come up again with the next flaw.

Question three: what did we do before the patch arrived?

ASD’s first alert went out before fixes were available. Between the alert and the release, organisations had a choice: wait, put temporary filtering in front of the site, change configuration, or take exposed systems offline. Tenable’s Gavin Millard summarised the logic in iTWire: “If you can’t patch it, protect it. If you can’t upgrade it, pull it down.”

Ask which of those your team chose, who approved it, and how long the gap was. Here is a simple calculation worth putting to management. If the first attempt was logged on 17 October and fixed versions were released at 1am AEDT on 22 October, any patch target measured from the release date started counting several days after attackers had begun probing. A dashboard that shows “patched within 48 hours of release” will look green and still miss the exposure window. Boards should ask for time exposed, not only time to patch.

Question four: have we looked for signs someone got in first?

This is where most organisations stop too early. Securonix’s field CISO, quoted by iTWire, said: “Do not let a successful patch close the ticket.” The recommended hunt covers suspicious POST or multipart requests, memory exhaustion, repeated PHP worker crashes and unexpected processes launched by PHP. Where code execution is suspected, the advice is to isolate and rebuild the system, then rotate database credentials and API secrets.

The credential point deserves emphasis. Wiz, one of the firms that found the flaws, said server memory exposed by the bugs “may contain passwords, authentication tokens, API keys and certificates”, as quoted by Information Age. A web server usually holds the password to the database behind it. If an attacker read memory before you patched, a fully updated server can still be serving an attacker who now holds working keys.

We made the same argument after ASD confirmed Australian organisations were compromised through Citrix NetScaler: patching closes the door, it does not tell you who walked through it. A targeted compromise check on internet-facing PHP systems, ideally with an independent web application test afterwards, is the proportionate response for most mid-sized organisations.

Question five: if we find something, are we ready for the obligations?

If the hunt does find evidence of access, the matter changes category. Under the Notifiable Data Breaches scheme, organisations covered by the Privacy Act must notify affected individuals and the OAIC when a data breach is likely to result in serious harm. A website with a customer database behind it is exactly where that assessment tends to land. APRA-regulated entities and critical infrastructure operators have their own notification duties on top.

The board’s question is whether the people who would make that assessment know they may need to, and whether legal and communications advisers are already on notice. It is far easier to stand a process down than to start one cold.

What to do first, this week

In order:

  1. Ask management for a written list of every system running PHP, including those run by suppliers, with versions before and after and the time each was fixed.
  2. Get written confirmation from hosting, SaaS and agency providers that their runtime is on a fixed version. Note anyone who cannot answer.
  3. Confirm workers and containers were restarted after patching, since an update that has not been loaded does nothing.
  4. Commission a compromise check on internet-facing PHP systems, and rotate database and API credentials wherever the check cannot rule out access.
  5. Write down who has authority to take a public service offline on suspicion, and what the trigger is. Put it in the incident response plan.

None of this needs a large program. It needs someone with enough seniority to ask the questions and enough independence to accept an uncomfortable answer.

How CISO Advisory Australia helps

We are independent. We do not resell hosting, patching tools or security products, so our only interest is whether your answers hold up. For a situation like PHP2Shell, that usually means one of three things: a short engagement to test management’s account of exposure and remediation and report it to the board in plain language; a compromise check and cyber audit of internet-facing systems; or an ongoing virtual CISO retainer, so the next critical alert lands with someone who already knows your environment and your suppliers.

Scope and fees are set out on our pricing page. A first conversation costs nothing and commits you to nothing.

The bigger lesson from this week is about speed. AI-assisted research turned a flaw that sat in PHP for years into a working exploit, and the people quoted this week expect more of the same. Organisations that know their assets, have pre-agreed decision rights and treat a patch as the start of an investigation will handle the next one in hours. Those that do not will be relying on luck.

If you want an independent senior view on how your organisation handled PHP2Shell, contact CISO Advisory Australia or call our 24/7 line on 07 2112 8502. We will tell you what we would ask, what a good answer looks like, and what to fix first.

Frequently asked questions

What is PHP2Shell?

PHP2Shell is the name given to two critical vulnerabilities in the PHP scripting language, CVE-2026-109618 and CVE-2026-109619. PHP says they can allow unauthenticated remote code execution on servers handling web requests. Fixed versions are PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35.

Does updating WordPress or Drupal fix PHP2Shell?

No. The flaws sit in the PHP runtime that the website software runs on, not in the content management system itself. The server's PHP version has to be updated, and the web server and PHP workers restarted so the fix takes effect.

Was anyone in Australia attacked through PHP2Shell?

ASD said it was not aware of successful active exploitation within Australia. Overseas, Cloudflare reported early attacks and Amazon logged an exploitation attempt on 17 October, before public disclosure. Organisations should check their own logs rather than assume they were not targeted.

We have patched. Is there anything else to do?

Yes. The bugs can expose server memory that may hold passwords, tokens and API keys, so patching does not undo any access that happened beforehand. Review logs for suspicious requests and unexpected processes, and rotate database and API credentials wherever you cannot rule out compromise.

Does PHP2Shell trigger a notifiable data breach?

Only if there is a data breach involving personal information that is likely to result in serious harm. The vulnerability alone is not a breach. If a compromise check finds evidence of access to systems holding personal information, the Notifiable Data Breaches scheme assessment applies and should start promptly.

Talk to a Virtual CISO

Need this handled for your organisation?

Confidential and no obligation. We respond the same business day — on-site same day / next business day, or remote, Australia-wide. Prefer to talk now? Call us 24/7 on 07 2112 8502.

Confidential. We typically respond same business day — or call us 24/7.

Frameworks & standards we assess and advise against

Independent, vendor-neutral expertise across the Australian and international frameworks government, regulators and boards rely on.

E8
Essential Eight
ISO
ISO/IEC 27001
NIST
NIST CSF 2.0
CPS
APRA CPS 234 / 230
ISM
ACSC ISM
PSPF
PSPF
IRAP
IRAP readiness
SOC2
SOC 2
PCI
PCI DSS
NDB
Privacy Act / NDB
SOCI
SOCI Act