⌕
News

OpenAI AI Model and Apple Security: What We Know About the 35 Vulnerabilities Claim

A viral claim says an OpenAI model found 35 Apple vulnerabilities. Apple's security credits confirm three WebKit findings attributed to OpenAI, not 35 device-takeover flaws.

f X wa in ↗
Illustrative cybersecurity image for the FBI ARTEX advisory story.
☷  Table of Contents
    Published:

    A claim that a new OpenAI model discovered 35 vulnerabilities in Apple devices — including flaws that could allow hackers to take control of iPhones and Macs — is circulating online. But the available evidence does not support the headline figure of 35 vulnerabilities being directly discovered by an OpenAI model.

    Apple’s published security acknowledgements provide a more specific picture: in its June 2026 security updates, Apple credited OpenAI’s Codex Security with identifying three WebKit vulnerabilities. The wider update addressed more than 30 security issues across Apple software, but that total should not be attributed entirely to OpenAI.

    What has OpenAI’s security tool actually found?

    According to Apple’s security acknowledgements, OpenAI Codex Security was credited for three vulnerabilities affecting WebKit, the browser engine used by Safari and other apps that display web content on Apple platforms.

    WebKit flaws can be serious because specially crafted web content may trigger unexpected behaviour or, in some cases, allow code execution. The impact depends on the specific vulnerability, the affected software version and the protections in place. Apple’s published credits do not substantiate the claim that OpenAI found 35 Apple vulnerabilities or that the three credited issues gave attackers unrestricted control of iPhones and Macs.

    Why the number 35 is misleading

    A report examining the viral claim found no confirmed record of an OpenAI model identifying exactly 35 vulnerabilities across Apple devices. The number appears to conflate the three WebKit issues credited to OpenAI with a broader Apple security update that fixed more than 30 flaws, with AI tools contributing to the wider security effort.

    Also Read  Karnataka Free Housing Sites Scheme 2026: 5 Lakh Sites for Landless Families

    It is important to distinguish between vulnerabilities found by one named tool and the total number of vulnerabilities addressed in a vendor’s security release. The two figures describe different things.

    OpenAI’s larger vulnerability research effort

    The Apple findings are part of a broader expansion in AI-assisted security research. In November 2026, OpenAI described an autonomous cybersecurity pipeline that identified 948 distinct Linux kernel local privilege-escalation vulnerabilities during an October research run. OpenAI said the pipeline also generated proof-of-concept exploits and patches for the issues.

    Linux kernel privilege-escalation vulnerabilities can allow an attacker who already has some level of access to gain greater control of a system. These findings concern the Linux kernel, not Apple devices, and should not be combined with the separate WebKit credits.

    Can hackers take control of iPhones and Macs?

    Some Apple vulnerabilities can potentially lead to arbitrary code execution or elevated privileges when successfully exploited. That does not mean every flaw gives an attacker full control of a device, or that the specific OpenAI-credited WebKit findings have been used in real-world attacks.

    Apple’s security advisories explain the impact and affected versions for each patched vulnerability. Users should rely on those advisories and install the latest compatible operating-system and browser updates rather than assume their device has been compromised because of a viral headline.

    What Apple users should do

    • Install software updates: Update iOS, iPadOS, macOS and Safari to the latest versions available for your device.
    • Enable automatic updates: Turn on automatic security updates where supported.
    • Be careful with links and attachments: Avoid opening unexpected links or files, particularly from unknown senders.
    • Use trusted app sources: Install apps through trusted channels and review app permissions.
    • Restart and update older devices: If a device no longer receives security updates, consider upgrading to a supported model for ongoing protection.
    Also Read  IRCTC Records Highest-Ever Single-Day Ticket Bookings on September 7

    AI is changing cybersecurity — for defenders and attackers

    AI tools can help researchers inspect large codebases, identify patterns associated with security flaws and develop fixes more quickly. The same general capabilities can also lower the effort required for malicious actors to search for weaknesses. That makes timely patching, responsible vulnerability disclosure and strong software security practices increasingly important.

    For now, the confirmed Apple-specific takeaway is narrower than the viral claim: Apple credited OpenAI Codex Security with three WebKit vulnerability findings in its June 2026 security acknowledgements. The claim that an OpenAI model found 35 Apple vulnerabilities, including flaws proven to let hackers take over iPhones and Macs, has not been substantiated by the cited records.

    Sources

    About the Author

    The InfoFlick editorial team covers the latest news, entertainment, technology, lifestyle, business, education and more.

    Related Stories

    View All →