AI Uncovers Hidden PHP Vulnerabilities, Patches Urged

Key Takeaways

• AI tools OpenAI GPT-5.6 Sol and Google Gemini 4 Argon helped discover PHP2Shell, two critical flaws affecting about 70% of websites
• The vulnerabilities had existed for over a decade and could let attackers remotely run code and steal data without passwords
• The PHP Foundation released emergency patches on Wednesday for versions 8.5.12, 8.4.27, 8.3.36 and 8.2.35
• The NSA, FBI, and governments worldwide urged immediate patching or taking vulnerable systems offline
• Wiz Research compared the severity to the 2014 Heartbleed security flaw
• Cohere released Rerank 5, supporting 100+ languages with 32K token context and calibrated relevance scores
• Dell reported a $95 billion AI backlog but faces 60-120 day delivery delays from component shortages
• SAP beat Q3 cloud revenue estimates with €6.67 billion, up 25 percent, driven by AI demand
• Denver and Denver International Airport lack AI governance frameworks, with no rules on data use or approved tools
• Amazon's AI shopping tool can auto-purchase items, and Visa is building payment safeguards for AI agents

AI finds critical flaw putting 70% of websites at risk

AI tools discovered two severe security flaws in PHP, a programming language that powers about 70% of the world's websites. The US National Security Agency urged organizations to take websites offline or install emergency fixes immediately. Security firms Calif and Wiz found the vulnerabilities, collectively called PHP2Shell, with help from OpenAI's GPT-5.6 Sol and Google's Gemini 4 Argon. Governments and companies worldwide scrambled to patch systems after the flaws were disclosed on Wednesday.

Global websites shut down after AI reveals PHP flaw

Government and company websites around the world went offline temporarily after AI helped uncover critical bugs in PHP. The vulnerabilities, called PHP2Shell, could let attackers hack websites and steal sensitive data like passwords. OpenAI, Google-owned Wiz, and Calif discovered the flaws, which had been hidden for over a decade. Dutch and Japanese leaders urged quick action, and companies like ANA, JR West, MUFG, and Wikipedia took precautionary steps.

AI discovers flaw that could hit 70% of websites

The FBI warned that vulnerabilities in PHP could let attackers remotely run code on websites. Governments and corporations raced overnight to update software after the AI-assisted discovery. PHP powers about 70% of websites worldwide, making the flaw a major concern. Many sites remained online, but some experienced outages while security teams worked on patches.

AI finds internet-wide flaw in PHP

Governments and companies are rushing to fix two critical flaws in PHP, a language behind 70% of the world's websites. The bugs, named PHP2Shell, could allow attackers to break into sites and read sensitive information. The flaws existed for about ten years before AI tools like GPT-5.6 Sol and Gemini 4 Argon helped researchers find and address them. The PHP Foundation released a patch and urged all website operators to update immediately.

AI reveals two-decade-old PHP security flaws

The NSA warned of two critical vulnerabilities in PHP that have existed since 2004. The flaws could let attackers access server file systems and connected databases. Calif used OpenAI's GPT-5.6 Sol and Wiz used Gemini 4 Argon and GPT-5.6 Sol to discover the bugs. Sites running unsupported PHP 8.1 or older are told to upgrade to version 8.5.12 as quickly as possible.

AI reveals critical PHP security flaws forcing websites offline worldwide

AI discovered critical security holes in PHP, the programming language that runs about 70 percent of the world's websites. Security researchers found the flaws in August using advanced AI models and kept the discovery secret while developers created a fix. The PHP Foundation released a patch on Wednesday and urged all users to update immediately. Many websites, including government sites and Pokemon apps, went offline temporarily as a precaution. Experts say no confirmed attacks have happened yet, but smaller websites may remain at risk for weeks.

PHP security emergency: AI uncovers decade-old web vulnerabilities

PHP, a scripting language used in web development, suddenly made headlines after two very high-risk security flaws were found. The PHP Group released an alert on October 20, 2026, and patches were issued for versions 8.5.12, 8.4.27, 8.3.36, and 8.2.35. AI helped researchers discover these vulnerabilities, which could have left over half the internet at risk. The flaws could be exploited much faster in the AI era than in the past. Website administrators are urged to verify their sites are running patched PHP versions.

Global websites go dark as AI exposes severe PHP security weaknesses

Government agencies and companies around the world shut down websites after two critical flaws were found in PHP. Wiz Research estimated PHP runs about 70 percent of the world's websites and called the flaws among the most severe ever discovered. The US National Security Agency told organizations to patch immediately or pull vulnerable systems offline. The flaws could let attackers take over web servers without needing a password. The PHP Foundation released an emergency update on Wednesday, comparing the severity to the 2014 Heartbleed flaw.

AI finds hidden PHP security bugs, prompting global website shutdowns

Artificial intelligence found critical security flaws in PHP, a programming language created in the early 1990s that runs an estimated 70 percent of the world's websites. Cutting-edge AI models from OpenAI and Google were used by security firm Wiz Research to uncover the hidden weaknesses. Security firm Calif also found the same flaws during an OpenAI-funded audit. Many websites, from the Paris police site to Pokemon apps, went offline as developers applied security updates. Experts say taking sites offline does not necessarily mean they have been hacked.

PHP2Shell flaws put millions of websites at risk after AI discovery

Two critical vulnerabilities called PHP2Shell, tracked as CVE-2026-109618 and CVE-2026-109619, were found in PHP's core engine. Wiz Research discovered them using AI models and announced the findings on October 21, 2026. The flaws could let attackers run code remotely and steal data without logging in. Affected PHP versions go back to PHP 5.4, meaning software more than 10 years old may be vulnerable. Google, OpenAI, AWS and other tech firms worked with researchers to analyze the flaws and build safeguards.

AI uncovers critical security flaws in PHP powering websites

Artificial intelligence helped find two critical security flaws in PHP, a programming language that powers many websites. The flaws involve how PHP handles computer memory and could let attackers run unauthorized code on web servers. Developers released fixes on Wednesday. Security firms Calif and Wiz discovered the vulnerabilities with help from AI models. Website operators must install the updates to stay protected.

AI finds decade-old PHP flaws causing global security scramble

Critical vulnerabilities in PHP were found using AI after remaining hidden for over ten years. About 2.6 million hosts use PHP, and roughly 1.5 million were running versions without a fix. Government sites in the Netherlands, Finland, Denmark and France took their services offline. Wikipedia, the Internet Archive and WordPress operator Automattic also addressed the problem. Security experts warned that attackers could mass compromise vulnerable web servers.

Governments scramble to fix AI-discovered PHP website flaws

Governments and corporations rushed to install security updates after two critical flaws in PHP were revealed. The FBI and UK government worked through the night on fixes. Some UK government services faced temporary outages. Experts said the flaws could let hackers control web servers without passwords. Japan advised website administrators to install patches immediately. One flaw corrupts system memory and another improperly stores user data.

Cyber experts use AI to uncover major website security flaw

Cybersecurity experts used artificial intelligence to uncover a major flaw affecting about 70% of websites. The vulnerability was found in PHP, a programming language widely used across the internet. The discovery prompted urgent security updates worldwide. The flaw was identified and reported to help protect web servers from potential attacks.

AI discovers critical flaw threatening 70% of websites

AI discovered a critical security flaw that threatens approximately 70% of websites. The vulnerability was found in PHP, a programming language that powers much of the internet. The discovery raised concerns about website security worldwide. The finding showed how AI tools can identify dangerous weaknesses in widely used software.

Afda and ZenRobot launch practical generative AI course for creatives

Afda, The No. 1 School for the Creative Economy, partnered with ZenRobot to launch an online generative AI course for content creators. The course teaches image creation, visual storytelling, motion, video and audio using AI tools. It includes four online sessions and costs R4,500 per person. The programme is designed for industry practitioners and postgraduate students.

Dell urges earlier AI hardware planning as component shortages cause delays

Dell Technologies is asking enterprise buyers to involve the company earlier in AI infrastructure projects. Component shortages are causing 60, 90 or 120 day delivery delays. Dell reported $95 billion in AI backlog at the end of its latest fiscal quarter. The company also faces shortages in DRAM, NAND, CPUs, disk drives and optical components.

NeoSoul to connect NeoTrade AI auto-trading with PredictFun

NeoSoul plans to integrate its AI auto-trading service NeoTrade with prediction market platform PredictFun through API integration. NeoTrade is currently in a paper-trading validation stage and does not use real funds. A live-trading feature will be released later after confirming service stability. NeoTrade will be the first AI auto-trading service integrated with PredictFun.

Cohere releases Rerank 5 with passage context and calibrated scores

Cohere AI released Rerank 5, a new reranker family available in Pro and Fast tiers. It supports over 100 languages with a 32K token context window and includes Context Focus, which returns relevant passages inside documents. Rerank 5 also offers calibrated relevance scores that read as probabilities. The model is generally available through integrations like LangChain and Elasticsearch.

Axios Live explores how AI agents are changing financial transactions

Axios held a live event in Las Vegas discussing how AI agents are changing banking, shopping and trading. FDIC chairman Travis Hill spoke about the need for thoughtful regulation of AI in finance. Alpaca co-founder Yoshi Yokokawa cautioned against fully automated trading and warned about risks when agents go rogue. The event was sponsored by J.P. Morgan.

Denver and airport AI oversight gaps found by auditors

Auditors found that Denver lacks strong rules for artificial intelligence. City employees can enter private data into public AI systems without alerts. Denver International Airport also has no AI governance framework. Airport staff use seven unapproved AI tools on the network. The city and airport agreed to follow all audit recommendations.

SAP cloud revenue beats estimates on AI demand

SAP reported third-quarter cloud revenue of €6.67 billion, up 25 percent. This beat the analyst estimate of €6.57 billion. CEO Christian Klein is leading the company's AI strategy. SAP shares rose more than 60 percent from the July low. The company plans to end standard support for older products next year.

AI shopping tools spend your money but who pays for mistakes

AI shopping assistants can find deals and buy products for consumers. KSL tested an Amazon AI tool that could purchase items when prices dropped. Banks like Bank of America and Capital One warn consumers to watch permissions given to AI. If AI buys the wrong item, consumers may be responsible for charges. Visa is building payment systems with limits for AI agents.

Sources

• Websites go dark after AI exposes programming language security flaw
• Some websites shut down globally as operators address PHP issue after AI finds critical flaw
• AI discovered a software vulnerability that could impact 70% of websites worldwide. The rush is on to patch it
• Websites shut down globally after AI finds flaw affecting 70% of the internet
• AI Tools Uncover Two-Decade-Old PHP Flaws as NSA Urges Urgent Patching
• Websites down: what is PHP and how has it raised AI hacking fears?
• AI helped uncover a security nightmare in the web’s backbone. Should you be worried?
• Websites go dark worldwide as AI exposes critical security flaws
• Websites down: what is PHP and how has it raised AI hacking fears?
• Decade-old PHP bug puts websites at risk: How AI uncovered it, who is vulnerable and how to stay safe | Explained
• AI Helped Uncover Critical Security Flaws in Software That Powers Websites. Here's What We Know
• AI uncovers decade-old vulnerabilities, leaving website operators scrambling
• Websites struggle after AI-discovered flaw sparks security scramble
• Cyber experts use AI to uncover website security flaw
• AI discovers critical flaw threatening 70% of websites
• Launch of practical generative AI course for content creators
• Dell Urges Earlier AI Hardware Planning Amid Component Shortages
• NeoSoul to Integrate AI Auto-Trading Service NeoTrade With PredictFun, Starting With Paper-Trading Validation
• Cohere Releases Rerank 5 With Passage-Level Context Focus and Calibrated Scores
• Axios Live: AI agents are changing how money moves
• Auditors find gaps in Denver and airport AI oversight
• SAP Cloud Revenue Beats Estimates on AI Push
• Your AI can shop for you. But who pays when it buys the wrong thing?