AI turned long-known PHP bugs into a mass exploitation crisis

Cloudflare logged millions of exploitation attempts within hours of the PHP2Shell bugs going public. AI models helped researchers at Calif, OpenAI and Wiz build working exploits for flaws known to the PHP team for years, though no successful breaches have been confirmed.

• Cloudflare saw millions of PHP2Shell exploitation attempts within hours
• Calif found the flaws with AI; OpenAI and Wiz independently confirmed them
• PHP Foundation reports no confirmed successful exploitation
• Microsoft patched a record 1,453 vulnerabilities this month