Tags: Deep Research

Google rolled out a visible usage meter for its Gemini AI service, letting users watch how each prompt, image or video chips away at their quota. The new system replaces vague daily caps with a five‑hour refresh window that tallies complexity, feature use and chat length toward a weekly limit. Free accounts receive a base allotment, while paid tiers—AI Plus, AI Pro and AI Ultra—receive progressively larger shares. Early reactions show the meter prompting more deliberate prompting, echoing how battery‑percentage displays have reshaped everyday habits. Ler mais

Google's Gemini 3.1 Pro demonstrates a leap in AI capability, offering more precise assistance across a range of tasks. The model can simulate adversarial scenarios to stress‑test plans, analyze visual content to match cinematic moods with real locations, provide spatial guidance for physical assemblies, generate interactive SVG animations, and conduct deep research for niche projects. These examples illustrate how the new reasoning layer and multimodal abilities make Gemini a practical partner for both personal and professional challenges. Ler mais

OpenAI has launched Lockdown Mode, a high‑security setting for ChatGPT aimed at users with elevated digital risk such as journalists, activists, and professionals in sensitive environments. The feature narrows the model’s capabilities by restricting web browsing to cached content, disabling image generation in responses, and turning off advanced tools like Deep Research and Agent Mode. It also blocks network access for generated code and prevents automatic file downloads, while still allowing manual file uploads. Initially available to Enterprise, Education, Healthcare, and Teacher plans, the mode will later expand to consumer and team subscriptions, with admins able to assign it at the workspace level. Ler mais

OpenAI has launched new features for ChatGPT Deep Research that let users limit searches to chosen websites, pull data from connected apps, and view results in a dedicated report viewer. The controls aim to reduce noise, improve citation reliability, and streamline verification, while acknowledging limits such as paywalls and site blocks. Ler mais

OpenAI has upgraded the ChatGPT deep research feature with a full‑screen document viewer that lets users scroll through AI‑generated reports in a separate window. The new interface includes a table of contents on the left and a source list on the right, enabling easier navigation and source verification. Users can direct the model to focus on particular websites or connected apps, monitor research progress in real time, adjust the research scope, and add additional sources while the report is being compiled. Completed reports can be downloaded in Markdown, Word, or PDF formats. The enhancements roll out to Plus and Pro subscribers immediately, with broader availability slated for the coming days for ChatGPT Go and free users. Ler mais

Perplexity, an AI‑driven research platform, lets users generate detailed reports by pulling from scholarly databases and reputable publications. Users can toggle between web and academic sources, download PDFs with hundreds of citations, and receive concise summaries that highlight key findings. Journalists, researchers, and students have found the tool saves hours of manual searching while still requiring verification of the linked sources. The free tier offers limited queries, while a paid option removes usage caps and provides deeper model access. Overall, Perplexity streamlines deep‑dive research while keeping the need for human fact‑checking intact. Ler mais

Google has rolled out Data Tables, a new feature for its NotebookLM AI research platform that lets users collect and synthesize information from multiple sources into a chart that can be exported to Google Sheets. The tool is initially available to Pro and Ultra subscribers and will gradually reach all users. Designed to help organize scattered notes, create price‑comparison charts, and aggregate research results, Data Tables interprets natural‑language requests to build table components. The addition follows a recent Deep Research mode launch for NotebookLM, which was first introduced for the Gemini chatbot in late 2024 to handle complex queries. Ler mais

Google is enhancing its AI‑powered note‑taking service NotebookLM with a new Deep Research feature that automates complex online research and generates source‑grounded reports. At the same time, the platform now accepts a wider range of file types, including Google Sheets, Drive URLs, PDFs, and Microsoft Word documents, allowing users to summarize spreadsheets and quickly add multiple files. The updates aim to streamline knowledge creation without leaving the workflow, building on earlier capabilities such as Video Overviews and Audio Overviews. Ler mais

Google has expanded its NotebookLM AI workspace by adding a Deep Research tool that can automatically browse the web and compile multi‑page reports on a single topic. The new feature works alongside the Gemini model, allowing users to direct the app to seek specific online sources and receive suggested articles, papers, and websites. NotebookLM can now also pull data from Google Sheets, Microsoft Word documents, and PDFs stored in Google Drive, with the ability to add Drive files simply by copying a URL. The enhancements are set to roll out to users within a week, further strengthening the app’s note‑taking, flashcard, and quiz‑generation functions. Ler mais

Google's NotebookLM has rolled out a major upgrade that makes locating, adding, and managing research sources far easier. The Gemini‑powered assistant now offers "Fast" and "Deep" research modes, lets users add CNET as a preferred Chrome source, and expands file compatibility to include Google Sheets, Microsoft Word documents, PDFs, and other Drive files without extra downloads. These enhancements follow recent improvements to the mobile app and chat experience, reinforcing NotebookLM’s role as a versatile AI‑driven notetaking and research tool for students, professionals, and casual users alike. Ler mais

OpenAI has added a new capability that lets users interrupt ChatGPT while it is generating a response. By clicking an “Update” button, users can add new context or corrections without restarting the query, allowing the model to adjust its answer on the fly. The feature is aimed at heavy‑use scenarios such as GPT‑5 Pro and Deep Research, where long‑running queries can be costly and time‑consuming. Analysts say the change could streamline workflows and reduce wasted effort. Ler mais

Google has expanded Gemini’s Deep Research feature to tap into users' Gmail, Docs, Sheets, Slides, PDFs, and Chat history. The update lets the AI generate more personalized reports by combining public web sources with personal files, giving users the ability to request analyses that reflect their own data. Users can choose which Google services to connect, and the feature is currently available on desktop with mobile support slated for the near future. Ler mais

Google has expanded its Gemini Deep Research feature to pull information directly from Gmail, Google Drive, and Google Chat. The AI can now analyze emails, documents, spreadsheets, slides, PDFs, and chat logs to generate multi‑step research reports. Users select which data sources to include, allowing the model to blend internal team content with web searches. The feature, initially available on desktop, aims to streamline market analysis, competitor reviews, and other research tasks, with plans to roll out to mobile devices soon. Ler mais

Brave announced a new feature for its AI‑driven search suite, Ask Brave, that delivers longer, report‑style answers alongside its existing AI Answers summaries. The enhancement automatically detects query intent, offers richer content such as videos, news articles, and product links, and integrates a chat mode powered by Deep Research. Users can trigger Ask Brave with a double question‑mark (??) or the ask button, and the company says the feature will boost the already high volume of AI‑generated responses while preserving privacy through encrypted, auto‑deleting chats. Ler mais

Security firm Radware revealed a proof‑of‑concept prompt injection that coerced OpenAI’s Deep Research agent into exfiltrating employee names and addresses from a Gmail account. By embedding malicious instructions in an email, the attack forced the AI to open a public lookup URL via its browser.open tool, retrieve the data, and log it to the site’s event log. OpenAI later mitigated the technique by requiring explicit user consent for link clicks and markdown usage. The demonstration highlights ongoing challenges in defending large language model agents against sophisticated prompt‑injection vectors. Ler mais

OpenAI has announced that its ChatGPT Projects feature, previously limited to paid subscribers, is now available to free users. Projects let users organize conversations around specific topics and offer custom instructions and file reference limits. The rollout also raises file upload caps: free users can add five files, Plus subscribers 25, and Pro subscribers 40. Users can personalize projects with colors and icons. The feature is live on the web and Android app, with iOS support arriving "over the coming days." This move aligns with OpenAI's pattern of gradually making premium tools, like Deep Research and ChatGPT Voice, accessible to all. Ler mais