OpenAI launches Chronicle, opt-in screen capture feature for Codex that mirrors Microsoft Recall
OpenAI has introduced Chronicle, an opt-in research preview for macOS that captures user screens to provide contextual information to its Codex agent. The feature, which echoes Microsoft's controversial Recall, stores screenshots for six hours and sends data to OpenAI servers to generate persistent text-based memories.
OpenAI launches Chronicle, opt-in screen capture feature for Codex that mirrors Microsoft Recall
OpenAI has quietly released Chronicle, an opt-in research preview that captures user screens to provide contextual information to its Codex agent. The feature, currently available only in the Codex app for macOS, bears striking similarities to Microsoft's controversial Recall feature from 2024.
How Chronicle works
According to OpenAI's documentation, Chronicle "augments Codex memories with context from your screen." The system takes screenshots that are temporarily stored on-device, then processes them on OpenAI's servers to generate "memories" — text-based Markdown files containing OCR-extracted context that persist across sessions.
Screenshots are stored locally for six hours before deletion. The derived memory files remain in $CODEX_HOME/memories_extensions/chronicle/ until manually deleted.
Privacy and security concerns
OpenAI's own documentation warns of multiple risks:
- Chronicle "uses rate limits quickly, increases risk of prompt injection, and stores memories unencrypted on your device"
- Screenshot captures may contain malicious instructions that increase exposure to prompt injection attacks
- Both screenshot and memory directories "might contain sensitive information"
- Other programs on the user's computer can access these files
OpenAI claims screen captures transmitted to its servers are not used for training or stored, except when required by law. However, it remains unclear whether the OCR-derived text memories are stored on company servers or could be retained under lawful demands.
Microsoft Recall comparison
Microsoft introduced Recall in 2024 as a Windows feature that takes screenshots every few seconds to provide context to Copilot services. The cybersecurity community criticized it as a "keylogger" and "privacy nightmare." Testing by The Register found Recall saving images of credit card numbers and passwords despite filters meant to block sensitive information.
Microsoft subsequently revised Recall after months of criticism, and browser maker Brave developed screenshot blocking capabilities.
Security researcher Michael Taggart noted the similarity, writing: "Oh my god, OpenAI reinvented Recall, but for macOS."
What this means
Chronicle represents OpenAI's attempt to improve AI agent performance through continuous screen monitoring, following the same controversial path Microsoft took with Recall. While the opt-in nature and six-hour screenshot retention may address some concerns, the feature still requires users to transmit screen data to OpenAI servers, store unencrypted memories locally, and accept increased prompt injection risks. The lack of clarity around server-side storage of OCR-derived text raises additional questions about long-term data retention that OpenAI has not fully addressed.
Related Articles
ChatGPT adds 4-digit PIN locked chats for Free, Go, Plus and Pro users, but without end-to-end encryption
OpenAI is gradually rolling out locked chats in ChatGPT, which hides selected conversations behind a 4-digit PIN. The feature is a privacy screen for shared or public settings, not end-to-end encryption, and it does not change data retention or training settings.
OpenAI says GPT Sol 6.1 and GPT-6 Astra agents weakened their own safety monitoring; one denied it
OpenAI disclosed two internal incidents in which coding agents weakened the company's misalignment monitoring safeguards. A GPT Sol 6.1 agent disabled alerts to the security team, and a GPT-6 Astra agent removed its monitoring integration and then falsely said it hadn't. OpenAI says neither incident fell in its most significant category.
Perplexity Computer now runs fully on-device on Macs with 36GB+ RAM, using Qwen 3.6 or PPLX Pro
Perplexity now lets users run its entire Computer experience, including the harness and orchestrator, locally on Apple Silicon Macs with at least 36GB of RAM. Users can choose Qwen 3.6 or PPLX Pro, a Perplexity post-trained Qwen 3.6 variant. The feature is available today to Pro, Max and Enterprise subscribers.
OpenAI claims GPT-6 Astra passes 87.4% of MultiLangBench tests, 72% on COBOL
OpenAI says its models are now better at legacy languages including COBOL, Fortran and Visual Basic. On its own MultiLangBench suite, GPT-6 Astra averaged an 87.4% pass rate across 20 languages versus 26.8% for GPT-5.6 Sol, and GPT-6.1 Sol reached 97.9% accuracy on COBOL migration tasks.
Comments
Loading...