The PHP project released fixes Wednesday for two critical flaws in the core engine of the PHP programming language, and governments and organisations took websites offline as a precaution while they patched, the BBC, The Guardian (opens in new tab) and The Verge (opens in new tab) reported. Tracked as CVE-2026-109618 and CVE-2026-109619 and known as “php2shell,” the use-after-free flaws could allow an unauthenticated attacker to execute arbitrary code on an affected system, Canada’s Centre for Cyber Security said.
Fixed releases are PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35; versions 8.1 and older no longer receive updates from the PHP project, the Canadian agency said. Security firm Wiz, which helped find the bugs, said every PHP version released in the last decade is affected by at least one of them, back to version 5.4, with severity reaching CVSS 9.2.
The flaws were found by researchers at Calif and Wiz using advanced AI tools; Calif and the PHP Foundation were part of an OpenAI program called Patch the Planet, the BBC reported. “Frontier models were critical to making this process practical, enabling us to develop and adapt working exploits across multiple applications in hours,” Wiz wrote. The BBC reported that details leaked on social media Tuesday, forcing the PHP Foundation to rush its announcement. Wiz estimates roughly 70% of the internet is affected, a figure the BBC said it could not verify.
Finland temporarily shut down its government’s public web services, and the Netherlands’ cyber-security agency took its own site offline, the BBC reported. In the U.K., 32 government services were taken offline, which cybersecurity minister Liz Lloyd told parliament “largely resulted from precautionary measures,” according to The Guardian (opens in new tab). The Wikimedia Foundation took down a number of systems, and the Internet Archive was briefly offline, the BBC and The Verge (opens in new tab) reported. The U.K.’s National Cyber Security Centre advised: “Where immediate remediation is not feasible, organisations should consider taking the system offline until it can be fully mitigated.”
The scale of exploitation is disputed. Cloudflare CEO Matthew Prince said the company saw tens of thousands of attempts in the 45 minutes after the announcement, rising to “hundreds of thousands within two hours, and millions now,” and Amazon Threat Intelligence said the flaw was “now being broadly exploited in the wild,” The Guardian (opens in new tab) reported. Ryan Austin, a Calif researcher, told The Guardian (opens in new tab): “At the moment it doesn’t look like exploitation has taken off or been widespread.” Canada’s Cyber Centre said in its Oct. 21 alert that no active exploitation had been identified at that time.
Cloudflare CEO Matthew Prince said Thursday that the number of exploitation attempts was now in the millions, NPR (opens in new tab) reported. The U.S. National Security Agency urged organizations to “maintain heightened awareness,” and the U.K.’s National Cyber Security Centre told them to “take immediate action,” according to NPR (opens in new tab).
“This is a significant week or two in cybersecurity,” Chris Krebs, former director of the U.S. Cybersecurity and Infrastructure Security Agency, told CNBC (opens in new tab) on Friday. He encouraged smaller companies to “constantly” scan for vulnerabilities, CNBC (opens in new tab) reported.
Update, Oct. 27: The U.S. Cybersecurity and Infrastructure Security Agency added both vulnerabilities to its Known Exploited Vulnerabilities catalog on Monday, Oct. 26, according to the agency’s notice, and encouraged all organizations to prioritize fixing the flaws listed in the catalog. The Register (opens in new tab) and BleepingComputer (opens in new tab) reported an Oct. 29 deadline for federal civilian agencies. The Register (opens in new tab) reported that the listing marks a shift from the week before, when the agency said it had not observed successful exploitation in the wild; SecurityWeek (opens in new tab) reported the same about CISA’s Oct. 23 alert. The Register (opens in new tab) said CISA has not disclosed which organizations were compromised and that the listing does not show how many production systems were breached.
SecurityWeek (opens in new tab) reported that Adel Karimi, a Perplexity researcher, recorded 35 shells launched across three honeypots on Oct. 24, from two IP addresses; The Register (opens in new tab) said the activity was consistent with exploitation of CVE-2026-109618 and that Karimi found no evidence the second flaw was involved. Karimi cautioned that the test used a favorable single-worker Apache mod_php setup: “The evidence demonstrates this chain on the intentionally favorable single-worker mod_php profile; it is not a measurement of exploit reliability on other worker counts, PHP builds, or applications,” he said, according to SecurityWeek (opens in new tab). No second-stage payload download was observed, SecurityWeek (opens in new tab) reported.
The Register (opens in new tab) reported that Delft University’s Threat Intelligence Lab has tracked more than 50,000 exploitation requests from over 600 IP addresses in 70 countries, and that 171 of those addresses went beyond probing to send exploit code or malware droppers. It also reported that Amazon spotted attempts as early as Oct. 17, four days before disclosure, and that Sophos walked back an earlier assessment after concluding the activity it saw was scanning and failed attempts, while Palo Alto Networks withdrew claims linking the flaws to an intrusion at a European government agency. BleepingComputer (opens in new tab) reported that on Friday the PHP security team shared a mitigation for a separate remote code execution flaw, CVE-2026-115560, that has no patch.