Quick Answer
ChatGPT for Mac users should update to version 26.924.20706 or later because OpenAI patched CVE-2026-100754, a flaw that could have let malware already running on a Mac take over the app and reach sensitive data. The bug was not a standalone remote attack, but updating now removes a weakness affecting chat logs and linked browser sessions.
Key Takeaways
- OpenAI fixed CVE-2026-100754 in ChatGPT for Mac version 26.924.20706.
- The vulnerability required malware to already be running on the targeted Mac.
- The reported exposure included ChatGPT chat logs, app-stored data, and linked browser sessions.
- The attack abused trusted script handling to bypass process-validation checks.
- ChatGPT for Mac users should install the latest available update and investigate signs of malware separately.
What is the ChatGPT Mac security flaw?
The ChatGPT Mac security flaw was a vulnerability tracked as CVE-2026-100754 that could have allowed malware on a targeted Mac to take over the ChatGPT application. OpenAI says its macOS security update 26.924 fixed the issue in ChatGPT for Mac version 26.924.20706 and credits Objective-See Foundation researcher Patrick Wardle for reporting it. OpenAI’s ChatGPT and Codex changelog identifies the patched version and the vulnerability identifier.
The flaw matters because ChatGPT for Mac can contain conversations and other data associated with the application. A malicious program that gains control of a trusted app can potentially access information that a user reasonably expects to remain inside that app’s normal security boundaries. The reported issue did not give an attacker a way to compromise an untouched Mac over the internet, but it could have made an existing malware infection more damaging.
ChatGPT for Mac users should treat the patch as a priority update, particularly if the Mac runs software from untrusted sources or shows signs of compromise. Users who want to limit retained conversation data should also review ChatGPT history and memory controls, although deleting old chats does not replace installing the security update.
How could malware abuse the ChatGPT for Mac vulnerability?
The ChatGPT for Mac vulnerability reportedly abused the application’s process-validation design rather than breaking its digital signature system directly. ChatGPT components verify digital signatures to determine whether requests come from trusted OpenAI software, according to technical details published by WIRED. The reported weakness appeared when a trusted script interpreter accepted an untrusted script.
The attack path reportedly allowed malware to launch the trusted interpreter repeatedly until the ChatGPT app’s parent and grandparent process checks were satisfied. Patrick Wardle told WIRED that his proof of concept required roughly a dozen lines of code. That detail is significant because a short proof of concept can make an implementation error easier for malware authors to adapt after a victim’s Mac is already infected.
| Attack element | Reported behavior | Why it mattered |
|---|---|---|
| Existing malware | Malware had to already be running on the targeted Mac. | The flaw increased the impact of an existing compromise instead of creating a standalone remote attack. |
| Trusted interpreter | A trusted script interpreter could accept an untrusted script. | Malicious code could appear to originate through a trusted process path. |
| Process checks | Launching the interpreter three times could satisfy parent and grandparent process checks. | The app could accept a request that its security design was intended to reject. |
| Patched version | OpenAI fixed CVE-2026-100754 in version 26.924.20706. | Updating removes the reported vulnerable behavior. |
The technical details and OpenAI’s response were described in WIRED’s report on the ChatGPT Mac vulnerability. The practical conclusion is straightforward: the update fixes one escalation route, but it cannot remove unrelated malware that may already be present on a Mac.
Who was affected by the ChatGPT Mac security flaw?
ChatGPT for Mac users running a version older than 26.924.20706 were the group affected by the reported vulnerability. OpenAI’s September 25, 2026, changelog says version 26.924.20706 fixed CVE-2026-100754, which means users who have installed that version or a newer release have the available fix.
The vulnerability required an attacker to have malware running on the Mac first. That condition limits the attack compared with a browser-based exploit or a malicious email attachment that compromises a computer without any prior foothold. At the same time, the requirement does not make the flaw unimportant, because malware infections often seek access to existing applications, accounts, and browser data after initial installation.
ChatGPT for Mac users who install apps only from trusted sources and keep macOS security protections enabled face a narrower practical risk than users whose systems are already compromised. The most sensible approach is to update ChatGPT promptly, then examine the Mac separately if there are unexplained login prompts, unfamiliar applications, or other indications that malware may already be active.
What data could the ChatGPT Mac flaw expose?
The ChatGPT Mac flaw could reportedly expose ChatGPT chat logs, other data stored by the application, and linked browser sessions after an attacker took over the app on a targeted Mac. Those categories matter because conversations can contain work context, personal details, copied text, files discussed in prompts, or references to other online accounts.
Linked browser sessions create a separate concern because an active session can sometimes provide access without requiring an attacker to know the account password again. The reported exposure does not establish that every affected user lost data or that every browser session would be reachable. The risk depended on a Mac already having malware and on the data and sessions available to the compromised application.
ChatGPT users should avoid placing passwords, financial account details, government identification numbers, and other highly sensitive information in AI prompts. A broader guide to ChatGPT privacy settings and data limits can help users reduce the amount of information retained in an account, but privacy settings cannot protect a device that malware has already compromised.
How do you update ChatGPT for Mac safely?
ChatGPT for Mac users should install the latest available app update and confirm that the installed version is 26.924.20706 or later. OpenAI identifies version 26.924.20706 as the release that fixed CVE-2026-100754, so users should not delay the update while waiting for additional public technical details.
- Quit ChatGPT for Mac so the application is not running during the update process.
- Open the app’s normal update mechanism or obtain the latest ChatGPT for Mac release from OpenAI’s official distribution channel.
- Install the available update and reopen ChatGPT for Mac after installation finishes.
- Check the application version and confirm that it is 26.924.20706 or a newer release.
- Review recent account activity and browser sessions if the Mac showed signs of malware before the update.
Independent update guidance also identifies version 26.924.20706 as the release that addresses CVE-2026-100754. The Hack Academy’s update guidance provides an additional summary of the patch and its purpose. The update fixes the reported ChatGPT weakness, but it does not substitute for investigating a suspected infection.
Mac users should stop and contact Apple Support, a trusted security professional, or an organizational IT team if the Mac appears actively compromised or if unfamiliar software cannot be removed safely. Do not rely on a ChatGPT update alone if malware is already running, because the underlying infection may affect other applications and accounts.
Does updating ChatGPT remove malware from a Mac?
Updating ChatGPT for Mac does not remove malware from a Mac because the OpenAI patch fixes the vulnerable application behavior rather than cleaning an infected system. The reported attack required malware to already be present, which means the app update closes one possible route for that malware but does not address the program that established access.
The distinction matters because a user could install version 26.924.20706 and still have an unrelated security problem on the device. Malware that reached a Mac through another route may still attempt to access browsers, files, saved passwords, or other installed applications. Users should therefore separate the two actions: update ChatGPT to remove the known flaw, then investigate the device if there is a reason to suspect compromise.
Mac users who believe malware has accessed chats or browser sessions should change important passwords from a known-safe device after checking account security pages. Users should also revoke unfamiliar sessions where supported. The practical response is to prioritize email, financial, work, and password-manager accounts because those accounts can affect other services if an attacker retains access.
What did OpenAI say about the ChatGPT Mac patch?
OpenAI said the company continues to evolve its security practices but recognizes that it needs to move faster, spokesperson Shane Bauer told WIRED. The statement followed public details about the vulnerability and OpenAI’s September 25 security update, which credited Patrick Wardle of Objective-See Foundation for identifying the issue.
The public timeline shows that the patch was available before the technical details were widely reported on October 2, 2026. That sequence gives users a practical opportunity to protect their systems before the reported proof-of-concept technique becomes more broadly understood. The available reporting does not describe a confirmed mass exploitation campaign, so users should focus on patching rather than assuming an account has been breached.
OpenAI’s patch is also separate from concerns involving AI agents and user data. Readers following reports about ChatGPT user images exposed by AI agents should treat that issue as a distinct incident with different circumstances. The immediate action for this vulnerability remains the same: update ChatGPT for Mac and investigate any existing malware risk independently.
FAQ
Does ChatGPT for Mac have a security update?
ChatGPT for Mac has a security update that fixes CVE-2026-100754 in version 26.924.20706. OpenAI listed the patch in its September 25, 2026, macOS security update changelog.
Was the ChatGPT Mac flaw a remote attack?
The ChatGPT Mac flaw was not described as a standalone remote attack. The reported attack required malware to already be running on the targeted Mac before it could abuse the vulnerable app behavior.
What could attackers access through the ChatGPT Mac flaw?
The ChatGPT Mac flaw could reportedly expose chat logs, other app-stored data, and linked browser sessions. The actual exposure depended on what data was available on a Mac that already had malware running.
Which ChatGPT for Mac version fixes CVE-2026-100754?
ChatGPT for Mac version 26.924.20706 fixes CVE-2026-100754, according to OpenAI’s changelog. Users should install that version or a newer available release.
Should I change my passwords after updating ChatGPT for Mac?
ChatGPT for Mac users should change important passwords from a safe device if they suspect malware was already active on the Mac. Updating the app removes the reported weakness, but it does not remove malware or invalidate a browser session that an attacker may have accessed.