All articles tagged with #chrome
WIRED highlights five Chromium-based extensions—Improve YouTube, Enhancer for YouTube, PocketTube, Turn Off the Lights, and DeArrow—that enhance YouTube with UI tweaks, playback controls, subscription organization, screen dimming to reduce distractions, and crowdsourced improvements to video titles and thumbnails, all without focusing on ad blockers.
Chrome on macOS can download a large local Gemini AI model (~4GB). To stop this, turn off On-device AI in Chrome via chrome://settings/system, and, if desired, delete the local model folder at ~/Library/Application Support/Google/Chrome/Default/OptGuideOnDeviceModel to reclaim space. The article also notes that disabling Apple Intelligence can free additional disk space and points to Google Support for more details.
Google Chrome reportedly rolled out a 4GB on-device AI model called Gemini Nano to eligible desktops between late April and early May 2026 without user consent or a prompt. The model runs locally (not via Google servers) and can be removed by toggling off On-device AI or uninstalling Chrome, with specific steps for Mac and Windows to detect and delete the files. Google says the model will auto-uninstall if resources are insufficient and has introduced a setting to disable or remove it; privacy advocates warn the rollout could violate GDPR and raise concerns about transparency and user control.
Google’s Magic Pointer, first shown with Googlebook, is rolling out to Chrome on desktop as Gemini in Chrome, letting you point at on-page elements and have AI infer context and act without lengthy prompts. DeepMind has published live AI Studio demos for image editing and map tasks to showcase the frictionless point-and-ask workflow, a stepping stone toward the OS-wide AI integration across desktop files and Android apps with Googlebook this fall.
Google’s AI-powered Magic Pointer, showcased with Googlebook, is rolling out to Google Chrome on Windows and macOS, letting Gemini interpret what you hover over to suggest actions (like comparing products or visualizing a couch). The feature expands beyond Googlebook devices, with Google stating you can use your pointer in Chrome to interact with pages more naturally, though rollout regions aren’t fully disclosed yet.
Google previews Android 17 with Gemini Intelligence, a broad on-device AI upgrade that can autonomously perform tasks, draw context from the screen, enhance web research and autofill, and add features like Create My Widget and screen reactions; rollout begins this summer on Galaxy and Pixel devices, with Gemini in Chrome arriving in late June and broader integration across wearables, cars, and laptops to follow.
Chrome has been quietly downloading a 4GB on-device AI model file named weights.bin to power Gemini Nano features, often after Chrome 148.0.7778.97, without user consent. To check on macOS, go to ~/Library/Application Support/Google/Chrome/Default/OptGuideOnDeviceModel/weights.bin and verify its size. To remove it permanently, disable On-device AI in Chrome Settings > System (this also stops future downloads); if the toggle isn’t present yet, use chrome://flags to disable AI-related flags and manually delete weights.bin. If you’re uncomfortable with this, consider switching to another browser.
Windows 11 introduces a registry policy called GenAILocalFoundationalModelSettings that lets admins block Google Chrome (and Edge) from automatically downloading the 4GB Gemini Nano on‑device AI model. Set GenAILocalFoundationalModelSettings to 1 under HKLM\SOFTWARE\Policies\Microsoft\Edge and HKLM\SOFTWARE\Policies\Google\Chrome to block the model (0 allows it). The policy supports dynamic refresh, works for Windows 11 Pro via Group Policy or registry edits, and is also available on macOS (not on iOS).
Chrome's 4GB on-device Gemini Nano AI model has been part of the browser for years, not a new download; its presence depends on hardware, account features, and whether you visit sites using the on-device API. Google says Chrome will delete the model if storage is low and you can disable local AI in Settings to avoid it. The confusion stems from Google's messaging—Chrome 148 tweaked wording about data handling—though Google insists on-device processing for non-Google sites and data sent to Google only when using Google sites. The bottom line: you should have the option to opt out, but AI is increasingly the default in Chrome.
Chrome quietly downloads a 4GB Gemini Nano model to support on-device AI features (the file is weights.bin in the OptGuideOnDeviceModel folder) when you opt in, without clear notification; you can disable On-device AI in Chrome Settings to reclaim space, and Google says the model will auto-uninstall if storage is low, with a recent option to turn off or remove it.
OpenAI today launched Codex for Chrome, a Chrome extension that lets Codex run directly in the browser on Macs and PCs, enabling it to test web apps, share context across tabs, use DevTools, and more without taking over the browser; the extension is installable via the Codex Plugins menu and adds to OpenAI's growing Codex user base of over 4 million weekly active users.
OpenAI released a Chrome extension for Codex that enables testing web apps, gathering context across open tabs, and parallel use of Chrome DevTools, broadening Codex's appeal beyond developers; the company plans eventual unification with ChatGPT and Atlas, and the extension works on Windows and Mac.
Chrome ships a 4-GB Gemini Nano on-device AI model. You can turn it off in Settings > System > On-device AI; deleting the file will be re-downloaded on reboot. Turning it off stops updates and may disable on-device scam detection and certain APIs, though Google says data stays on-device when enabled. If you’re uncomfortable with it, you can simply use a different browser.
A 4GB file named weights.bin has surfaced in Chrome folders, tied to Gemini Nano on-device AI that runs features like autofill and text summaries; Google says the model is downloaded to keep on-device features ready and has added an opt-out, but concerns remain about transparency and consent as users may not be clearly informed of the storage impact. Deleting it may not be permanent if related AI features stay enabled.
Google overhauls its Android and Chrome vulnerability rewards programs, elevating top Android payouts to $1.5 million for zero-click full-chain exploits on the Pixel Titan M2 (with up to $750,000 for non-persistent variants) and offering up to $250,000 for Chrome full-chain exploits plus a $250,128 bonus for MiraclePtr-protected memory allocations; the changes push for concise bug proofs instead of lengthy analyses and narrow Android focus to Linux kernel vulnerabilities in Google components unless researchers demonstrate device exploitability. The revamp follows a record 2025 with $17.1 million paid to 747 researchers, bringing total payouts since 2010 over $81.6 million, and 2026 totals are expected to rise despite some reductions.