OpenAI Adds Chronicle to Codex for Screen Monitoring
OpenAI launched Chronicle, a capability within its Codex application. This tool captures screen activity to generate memories. Codex draws on these memories to provide context for upcoming work. Users benefit from reduced need to repeat details about references, tools, or projects.
Chronicle operates quietly in the background. AI processes the recordings and produces summaries. These summaries save as Markdown files on the local device. OpenAI states the actual recordings remain short-term. They delete automatically after six hours.
Feature Rollout and Access
The rollout starts as an opt-in preview. It targets ChatGPT Pro subscribers using macOS. Residents in the EU, UK, or Switzerland cannot access it yet. To activate, users visit Codex settings. They select the "Personalization" section. From there, they enable "Memories" and then "Chronicle." macOS requires permissions for screen recording and accessibility features.
OpenAI founded in 2015 as a research organization focused on artificial general intelligence. Key figures include Sam Altman as CEO. The company gained prominence with the GPT language models. ChatGPT debuted publicly in late 2022 and quickly amassed millions of users. Subscription tiers like ChatGPT Pro offer advanced access to models and features. Codex traces roots to OpenAI's code-focused models from 2021. Those powered tools like GitHub Copilot for software development. The Codex app extends this into a dedicated interface for coding assistance.
Technical Details and Warnings
Chronicle processes visual data from the screen. This allows the AI to understand ongoing work without constant user input. Summaries store locally, aiding privacy to some extent. However, OpenAI highlights several concerns. The feature consumes rate limits at a fast pace. Users may hit caps sooner during heavy use.
Stay ahead of the AI curve
The most important updates, news, and content — delivered weekly.
No spam. Unsubscribe anytime.
Another issue involves security. Prompt injection attacks pose a threat. These occur when harmful instructions appear on websites shown on screen. The AI might incorporate them unintendedly. Memories save unencrypted on the device. This means local access could expose content without extra protection.
OpenAI continues to build memory functions across products. Earlier updates to ChatGPT included persistent memory for conversations. Chronicle pushes this further by integrating real-time screen data. Developers often seek such context awareness in AI coding helpers. It matches code snippets to project states or tool interfaces automatically.
ChatGPT Pro provides priority access to previews like this. Subscribers pay for enhanced capabilities, including longer context windows and specialized models. macOS support aligns with OpenAI's focus on desktop apps. The company released dedicated apps for Mac and Windows in recent years. Exclusions in Europe stem from regulatory hurdles like GDPR. OpenAI faces ongoing scrutiny there over data practices.
Broader Context in AI Development
Screen observation represents a step toward more autonomous AI agents. OpenAI invests heavily in agent systems that handle complex workflows. Chronicle fits into efforts to make tools proactive. Users enable it deliberately, balancing utility with control.
The feature arrives amid rapid AI advancements. OpenAI competes with rivals like Anthropic and Google in code generation. Codex models excel at translating natural language to code across dozens of programming languages. Background processing keeps performance smooth, though resource use remains a factor.
OpenAI draws from its source announcement for these details. Early adopters can test Chronicle now on supported systems.
Related on Neura Market