What happened
PHP2Shell Live’s report, published on 29 October and updated through 8 November, gives a later snapshot from TU Delft IntelliLab’s tracker. By 30 October at 11:12 UTC, the tracker had recorded 76,153 PHP2Shell-related requests from 680 source IPs across 317 autonomous systems and 73 countries. Its observation window closed at 11:11 UTC. The tracker counted 177 IPs that had sent code or a dropper. These are requests and source addresses observed by the tracker, not confirmed victims or compromised organisations.
The report also describes how the tracker classified requests by their furthest observed stage: 52,990 probes from 553 IPs, 10,510 code-execution requests from 148 IPs, 4,523 overload requests from 132 IPs, and 4,656 dropper requests from 50 IPs. It cautions that the categories overlap and should not be added together. Peak activity was 8,179 requests in the hour beginning 26 October at 14:00 UTC. A request classified as code execution or dropper delivery is stronger evidence of attack traffic than a probe, but it does not show that a production system was breached or data was taken.
The report places the telemetry alongside patch-uptake estimates. Citing LACNIC’s analysis of Censys data from 26 October, it says 12.8% of exposed PHP services with identifiable versions in Latin America and the Caribbean were patched, compared with 25.1% globally. The analysis covered 21 countries. LACNIC cautioned that the unpatched share includes PHP branches that no longer receive security fixes; the figures therefore do not say how much of the gap can be closed by patching rather than migration.
A separate survey reported by Boannews used Shodan data to count 1,773 patched servers among about 6,500 patch-eligible Korean PHP 8.2–8.5 servers, or 27.3%. Across all Korean PHP servers, including end-of-life branches, its estimate was about 5.5%. The report says the survey used ten queries and excluded servers whose versions could not be identified. The two patch estimates cover different populations and methods, so they should not be treated as directly comparable measures of a single global remediation rate.
What the archive adds
New Runtime’s earlier coverage, Delft honeypot data shows 72,000+ PHP2Shell exploitation requests from 659 IPs, with Mirai-derived droppers, dated 28 October, reported an earlier Delft snapshot of 72,343 requests from 659 IPs. This is a new snapshot on a known exploitation-telemetry story: the later count is higher, while the central caveat remains that request volume is not a victim count.
Trend signal
This update is consistent with the expectation that PHP2Shell exploitation will continue while unpatched internet-facing PHP hosts remain, stated in New Runtime’s earlier analysis. The later tracker snapshot supports continued attack activity, but does not establish successful intrusions or validate the scale of unpatched hosts. It is a directional signal, not proof of the assumption’s full scope.
Why now
The tracker’s observation window closed on 30 October, extending its earlier late-October snapshot. That timing makes the figure useful as a near-term measure of continued traffic after disclosure, but it does not make the count a measure of successful compromise.
Connections to other events
The later snapshot follows the earlier Delft measurement covered by New Runtime on 28 October. The two figures share the same important interpretive boundary: both count requests and source IPs observed by the tracker, not confirmed victims. The report also places the traffic beside patch-uptake estimates from LACNIC and a Korean Shodan survey, which illustrate that measured remediation varies by population and method.
Why it matters
For security and platform teams, attack telemetry can show that a vulnerability remains a live operational concern, while still being insufficient to answer the incident-response question: was our system compromised? Treating requests as victims risks overstating impact; treating lack of confirmed victims as proof of safety is equally unsound.
The patch figures also point to a version-lifecycle problem. Where services run unsupported PHP branches, a security fix may require a migration rather than a routine package update. Teams need an inventory that includes embedded and internally managed PHP, a way to distinguish supported from end-of-life versions, and checks that confirm the running build after remediation. The published aggregates can motivate that work, but cannot identify an organisation’s own exposure or establish its patch status.
Limits and uncertainty
The Delft count is tracker telemetry, not a census of vulnerable systems or a record of successful intrusions. The report does not establish whether each request reached a real, production PHP deployment, whether execution succeeded, or whether any data was accessed. Source IP counts also do not map directly to unique attackers or organisations.
Patch estimates have their own limits: LACNIC’s figures include end-of-life branches and depend on identifiable versions, while Boannews’ survey covers Korean servers found through ten Shodan queries and excludes servers with unknown versions. The estimates describe different samples and dates. None establishes the current status of a particular operator’s systems.
Reader takeaway
The 76,153 figure is evidence of sustained PHP2Shell-related attack traffic, not proof of 76,153 successful attacks—or any particular number of victims. Use it as a reason to verify your own PHP inventory and remediation status, not as a substitute for those checks or for incident evidence.
Open question
How much of the observed attack traffic reached vulnerable production systems, and how many organisations can independently confirm compromise?