Apple reportedly said that it is using Claude and OpenAI Codex Security to find flaws in macOS and WebKit.
Apple reportedly said that it is using Claude and OpenAI Codex Security to find flaws in macOS and WebKit
Photo Credit: Apple
Artificial intelligence is helping security researchers discover software vulnerabilities, and this trend appears to be creating a new challenge for Apple. The iPhone maker is now reportedly limiting the number of active security reports each researcher can have open simultaneously after a series of AI-assisted submissions have created a new challenge for the company. The AI tools are producing both genuine and false Mac vulnerabilities. Apple is said to be using AI to help prioritise incoming reports, but human experts still review them.
As reported by the Financial Times, AI-generated bug reports have prompted Apple to limit security submissions. The publication, citing cybersecurity firm Bynario, states that it discovered more than 50 potential security flaws in Apple's macOS operating system within three weeks, including a privilege escalation chain that could allow attackers to take full control of a Mac.
Bynario, using its Atlas security platform powered by GPT-5.5, reportedly uncovered a flaw in macOS Screen Sharing. The vulnerability allowed an authenticated Virtual Network Computing (VNC) client to access protected data and create files with root privileges under certain conditions. This attack affected systems in which Screen Sharing or Remote Management were enabled together with legacy VNC password authentication. Apple later assigned the flaw as CVE-2026-43760 and addressed it in its macOS Tahoe 26.6 update.
Bynario has reportedly developed a working exploit that demonstrated how the flaw could be used to execute commands with root privileges. It reportedly submitted the exploit to Apple, helping its engineers reproduce the issue and validate a fix.
These latest findings show AI's increasing role in uncovering vulnerabilities in less time. The Cupertino-based company reportedly said that researchers are using Claude and OpenAI Codex Security to find flaws in macOS and WebKit. The AI speeds up vulnerability discovery, but Apple still needs to manually verify each report before issuing a fix.
Apple has reportedly revamped its bug bounty programme to speed up this verification process with rewards of over $5 million (roughly Rs. 47 crore) for the most critical exploit chains. The programme also includes Target Flags to help researchers provide stronger proof of exploits.
Get your daily dose of tech news, reviews, and insights, in under 80 characters on Gadgets 360 Turbo. Connect with fellow tech lovers on our Forum. Follow us on X, Facebook, WhatsApp, Threads and Google News for instant updates. Catch all the action on our YouTube channel.