- Apple released security updates for iOS, macOS, and Safari to address over three dozen flaws, including four vulnerabilities in WebKit discovered using AI tools like Anthropic Claude and OpenAI Codex Security.
- The patched WebKit vulnerabilities include: CVE-2026-43707 (memory corruption), CVE-2026-43716 (unspecified issue), CVE-2026-43745 (out-of-bounds write), and CVE-2026-43715 (use-after-free).
- Apple credited the first three security defects to OpenAI Codex Security and acknowledged Milad Nasr and Nicholas Carlini from Anthropic for CVE-2026-43715.
- Apple stated it is making security updates earlier due to the potential of AI tools to accelerate exploit development and cyber warfare, reducing the time between discovery and weaponization.
Apple has patched over 30 vulnerabilities in its iOS, macOS, and Safari platforms, including four critical WebKit flaws discovered using AI tools like Anthropic Claude and OpenAI Codex Security.1345
The vulnerabilities include CVE-2026-43707, a memory corruption issue that could lead to unexpected crashes when processing malicious web content, which was addressed with improved memory handling.2

Apple has credited the first three security defects to OpenAI Codex Security, while the fourth, CVE-2026-43715, was acknowledged to researchers Milad Nasr and Nicholas Carlini from Anthropic.
In a statement to Reuters, Apple emphasized the urgency of these updates, stating, "it was adapting to the reality that, given the ability of artificial intelligence to speed the development of malicious hacking tools, it needed to reduce the time between when updates were first made public and when they were put into customers' hands." This proactive approach aims to mitigate the risks posed by AI-enabled cyber threats, which can shrink the window between vulnerability discovery and exploitation to mere hours.
“Apple has released security updates for iOS, macOS, and Safari to address over three dozen flaws. The updates include four WebKit vulnerabilities discovered using AI tools like Anthropic Claude and OpenAI Codex Security.”
