OpenAI ChatGPT Computer History Feature on macOS Could Expose Detailed User Activity Logs to Infostealers

OpenAI has introduced the Computer History feature in its macOS ChatGPT app, which records application switches, clicks, keystrokes, and accessibility context to generate AI summaries and memories. The feature is disabled by default and requires explicit activation of Memories, with availability limited to Pro, Business, and Enterprise users outside the EEA, Switzerland, and the UK. While raw event files are deleted after 48 hours and not used for model training, the resulting Markdown memory files remain unencrypted on the local Mac. These files can be read by any process running under the same user account, creating a ready-made activity log for infostealers and other malware. OpenAI also warns about prompt injection risks where hidden instructions from websites or apps could influence ChatGPT or Codex behavior. Users retain controls to select participating apps, pause collection, or delete history, but the lack of encryption on stored memories raises significant privacy concerns.

AntiMalwareAugust 19, 2026Translated from Russian

Read at source

OpenAI has rolled out a new Computer History feature in the ChatGPT desktop application for macOS that converts recent user activity into structured summaries and memories for use by ChatGPT and Codex.

The system captures application switches, mouse clicks, text input, keyboard shortcuts, and additional context obtained through macOS accessibility APIs. These raw events are processed locally for up to 48 hours before deletion, after which brief summaries are generated and stored as ordinary Markdown files on the device.

The feature is turned off by default and requires users to enable Memories. It is currently available only to Pro, Business, and Enterprise subscribers, with administrator approval needed for corporate plans. Access is blocked in the European Economic Area, Switzerland, and the United Kingdom.

Users can choose which applications and websites participate, pause data collection at any time, or delete records for the last ten minutes, hour, day, or entirely. Screenshots, audio, and private browsing activity are explicitly excluded from the logs.

Although OpenAI states that raw events are not retained on its servers and are not used for model training, the generated Markdown memory files remain unencrypted. Any other application running under the same macOS user account can therefore read them, giving infostealers a ready-made daily activity report.

OpenAI additionally cautions about prompt injection risks: hidden instructions embedded on websites or inside applications may enter the context window and cause ChatGPT or Codex to follow attacker-controlled directives.

OpenAI ChatGPT Codex macOS Computer History

ShareX Facebook LinkedIn Reddit

Meridian digestThe week's most consequential security stories, distilled into one briefing.See the digest

More in Privacy & Surveillance

Privacy & Surveillance

Maximizing iPhone Privacy: Essential Settings to Enable, Disable, and Verify Immediately

The article provides a detailed guide on strengthening iPhone privacy by limiting unauthorized access to messages, location data, photos, and cloud backups. It explains how to set a strong alphanumeric passcode, enable Stolen Device Protection, and restrict lock screen notifications and widgets. Users are advised to review Apple ID security, enable Advanced Data Protection for iCloud, and minimize app permissions for location, contacts, and camera. Additional steps include turning off cross-app tracking, using private Wi-Fi addresses, and activating Lockdown Mode for high-risk scenarios. The guide emphasizes regular updates, encrypted local backups, and periodic audits of shared access and active sessions to maintain long-term protection.

SecuritylabOct 28, 2026Translated from Russian

Privacy & Surveillance

Top 10 Secure DNS Servers for Privacy, Threat Blocking and Ad Filtering in 2026

Public DNS resolvers differ significantly in their ability to block malicious domains, filter advertising and tracking, respect user privacy, and support modern encryption protocols. The guide evaluates ten services including Quad9, Cloudflare, AdGuard DNS, NextDNS, Control D, CleanBrowsing, Mullvad DNS, DNS4EU, Google Public DNS and DNS.SB, providing IPv4 addresses, DoH and DoT endpoints, and recommended use cases. Key distinctions include Quad9’s focus on phishing and malware blocking without logging, Mullvad’s strict no-logs policy requiring encrypted protocols only, and DNS4EU’s EU-based infrastructure with mandatory court-ordered blocks. The article explains why DNSSEC and encryption alone are insufficient and outlines four practical selection criteria: threat filtering quality, data-handling practices, encryption support, and network stability. Detailed configuration instructions cover Windows 11, Android, iOS, and home routers, along with verification methods and warnings against mixing providers with conflicting policies.

SecuritylabOct 28, 2026Translated from Russian

Privacy & Surveillance

WebRTC IP Leaks Explained: Why Disabling the Protocol Harms Anonymity in Anti-Detect Browsers

The article from Aurorium details how WebRTC technology can leak real IP addresses even when users route traffic through proxies. It explains the peer-to-peer mechanics of WebRTC, the role of STUN servers in generating ICE candidates, and why standard TCP proxies fail to handle UDP traffic. Disabling WebRTC via flags or JavaScript extensions creates detectable anomalies that anti-fraud systems flag as non-standard environments. The recommended approach involves native emulation inside the browser kernel so that STUN requests return proxy IPs while preserving full API integrity across workers. Additional leaks through local network addresses, mDNS candidates, and IPv6 are also covered with practical verification methods using tools like BrowserLeaks and CreepJS.

HabrOct 26, 2026Translated from Russian