Websites pulled offline worldwide as AI finds critical security flaws

Government agencies and companies around the world shut down websites on Wednesday after two critical flaws came to light involving PHP, a programming language that is the basis for a large share of the internet.

PHP sits underneath popular web design platforms, so the weaknesses potentially affect a vast range of websites, from online stores and news outlets to government services.

Wiz, one of the security firms that found the issues, estimated that PHP runs about 70 per cent of the world’s websites and said the flaws rank “among the most severe vulnerabilities discovered in PHP and arguably in software history”.

The US National Security Agency told organisations to patch immediately or pull vulnerable systems offline. The flaws “often provide a direct path to the server’s file system or database”, the NSA said in a statement on social media.

Attackers could use them to take over web servers without needing a password.

Volker Dusch, a security engineer at the PHP Foundation, told The Wall Street Journal it was the most serious problem PHP had faced in more than a decade, comparing it to Heartbleed, a 2014 flaw that set off a worldwide rush to patch.

The discovery came amid a surge in software security disclosures driven by increasingly powerful AI models, with the number of vulnerabilities found in 2026 on track to roughly double last year’s record total.

PHP volunteers released a fix on Wednesday.

“If you run PHP, it is imperative that you update now,” the PHP Foundation, which funds developers working on the language, said in a blog post.

Security firm Calif found the flaws in August using OpenAI’s GPT-5.6 Sol model, as part of an OpenAI-funded audit of PHP’s code.

Google-owned Wiz said it found them independently weeks later with help from AI models including Google’s Gemini 4 Argon and GPT-5.6 Sol.

OpenAI said on social media that it considered the bug “one of the most impactful vulnerabilities in recent memory”, adding that it “marks a new phase, where AI can find security bugs with internet-scale impact.”

Australia slams OpenAI breach of government portal

Ethan Mollick, a professor at the University of Pennsylvania’s Wharton School who studies AI, said that companies he talks to are struggling to keep up with the pace of patches needed to shield from AI-driven cyberattacks.

“Now consumers are going to have the same experience of getting everything updated often and fast,” he said on social media.

The flaws had been reported privately to PHP’s developers and kept confidential while a fix was prepared.

Details leaked online on Tuesday, a day before the planned release of the fix, prompting the rush to take sites offline.

In France, local government websites, including one for Paris police, have been inaccessible since Wednesday afternoon at the request of the French National Cybersecurity Agency, the interior ministry said.

In the Netherlands, the national digital identity service DigiD, the website of the House of Representatives and the site of the national cybersecurity centre went offline as a precaution, according to Dutch public broadcaster NOS.

Belgium’s foreign ministry temporarily took its websites offline as a precaution, it said on social media, although the Centre for Cybersecurity Belgium had recorded no attacks on Belgian sites.

In Australia, the Australian Signals Directorate said its own cyber.gov.au site was among government websites affected by the flaws, according to public broadcaster ABC.