Full Summary
This Saturday, June 6th, OpenAI rolls out significant updates to ChatGPT, focusing on both enhanced utility and robust security. Multiple sources, including Tech Times, NewsBytes, The Hacker News, and Digital Trends, confirm the launch of "Lockdown Mode" across all ChatGPT account tiers, including Free, Plus, Pro, and Business. This optional security setting is designed to protect users from "prompt injection attacks" by severely limiting ChatGPT's ability to connect to the web or external services. When activated, features like live web browsing, image retrieval, deep research, and agent mode are disabled, although DALL-E image generation still functions. While it doesn't prevent malicious instructions from entering the model, it aims to block sensitive data from leaving the system, making it crucial for users handling confidential information. OpenAI also introduces an active session manager, allowing users to monitor and revoke access from connected devices. Concurrently, Moneycontrol.com, Gadgetbridge.com, and Connectedtoindia.com report a major "Dreaming" memory upgrade for ChatGPT. This new system allows the AI to remember user preferences and past conversations with greater precision, providing a more personalized and consistent experience. This "dreaming" architecture continuously processes and stores information in the background, improving factual recall and user preference following, with significant performance gains reported. Users gain control over this memory through a new summary page, where they can view, edit, or delete stored information. This memory upgrade is currently rolling out to Plus and Pro users in the United States, with wider expansion planned. In terms of new capabilities, the Jamaica Observer highlights ChatGPT's direct integration into Google Sheets, transforming it into a financial co-pilot for tasks like building budgets or cleaning data. Separately, Mint reports a new ChatGPT for PowerPoint add-in, enabling users to create, rewrite, and restructure presentations from source material in minutes. Finally, Lapaas Voice and TheWinCentral.com confirm that ChatGPT's web version now allows users to send emails directly from conversations, integrating with services like Gmail and Outlook, complete with a "Consent Card" for explicit approval before sending. These updates mean your digital workflows become more streamlined and personalized, but also that you now have more control over the security and privacy of your sensitive data when interacting with AI tools.