Gemini App Unveils Prompt Bar Redesign, Bringing Model Switching Mid-Conversation

Google is rolling out a significant overhaul to the user interface of the Gemini app on both Android and iOS, transforming the familiar prompt bar into a dynamic "bottom sheet." This redesign marks one of the most substantial visual and functional updates to the AI assistant’s mobile experience since its launch, aiming for greater efficiency and a cleaner aesthetic.

The traditional rounded rectangular prompt box is now completely gone, replaced by a bottom sheet that visually floats above the application’s main content. This new UI layer is designed to merge seamlessly with the device's open keyboard, using a subtle shadow effect along the top edge to distinguish it from the conversation history.

This transition from a static box to a dynamic sheet aligns the mobile app with modern design paradigms seen across other Google applications. By removing the container, the interface feels less cluttered, creating more horizontal screen real estate for users to type their queries.

Perhaps the most crucial functional update is the relocation of the AI model selector. Previously, users had to tap a dropdown menu near the top of the home screen to switch between models like Gemini 2.5 Flash and 2.5 Pro.

With the new design, the model picker is now embedded directly in the bottom-right corner of the prompt sheet, adjacent to a newly-circular microphone icon. This change streamlines the workflow, finally allowing users to select a different model for their next prompt without navigating away from the active chat.

Furthermore, Google has utilized this redesign to implement several user experience (UX) quality-of-life improvements. The '+' menu for adding inputs and the Gemini Live button are now positioned closer to the edges of the screen, benefiting from the freed-up space.

The standard but often disruptive disclaimer, "Gemini can make mistakes, so double-check it," has also been repositioned. Instead of sitting permanently below the prompt bar, it now appears concisely at the end of the AI's response, ensuring a cleaner visual flow while users are chatting.

The rollout for the box-less prompt bar redesign is now widely available for Gemini users on both major mobile operating systems. However, the accompanying change—the repositioning of the model picker—is being implemented gradually and is not yet universal across all accounts.

Also Read:

• Google Gemini Now Lets You Share Custom 'Gems'.
• Google Redesigns Gemini's Tools Menu for a Cleaner Mobile Experience.

Gboard Rolls Out 'Tap-to-Undo' Feature to Fix Annoying Auto-Corrections.

In a highly welcomed quality-of-life update, Google’s popular virtual keyboard, Gboard, is introducing a simple yet crucial feature that addresses one of the most frustrating aspects of mobile typing: aggressive auto-correction. Users can now easily spot and revert unwanted changes with a new tap-to-undo mechanism.

The new functionality works by subtly highlighting any text that Gboard has automatically changed while a user is typing a sentence. This visual cue makes it significantly easier for users to identify exactly which words were altered without needing to carefully proofread the entire text.

Once an auto-corrected word is highlighted, tapping on it immediately reveals the original, user-typed word as a suggestion in the prediction bar. This allows for a swift, single-tap reversal of the correction, restoring the text to the user's intended spelling or phrasing.

This method offers a marked improvement over previous correction methods, such as the "undo auto-correct on backspace" feature. That older tool often mistakenly reverted valid corrections when a user was merely attempting to fix a spacing error or a minor typo nearby.

While Gboard recently added dedicated undo/redo buttons, the tap-to-undo feature is a more elegant, context-aware solution. It saves valuable space on the toolbar and allows users to fix auto-corrections without interrupting the flow of their typing process.

The feature is currently rolling out to users, appears to be controlled by the main auto-correction setting, and is expected to be particularly useful when interacting with Gboard’s growing suite of AI writing and grammar-proofing tools. This enhancement reinforces Gboard’s status as a leading mobile keyboard by prioritizing user control and accuracy.

Google Meet Finally Adds Meeting Timers to Combat Endless Virtual Calls.

In a move aimed squarely at boosting productivity and curbing the scourge of sprawling virtual meetings, Google Meet has begun rolling out a highly anticipated native timer feature. This addition brings a visible, ticking clock into the meeting interface, creating a subtle but effective sense of urgency for all participants.

This new tool directly responds to the prevalent issue of "meeting creep," especially in organizations with distributed workforces. By providing a clear visual countdown, Google is offering users a simple mechanism to stick to their intended agenda and time limits.

The feature is accessible via the Meeting tools button, located in the lower right corner of the desktop user interface. Depending on the meeting's settings, the timer can be set and controlled by either the meeting host or all participants.

Users have the flexibility to set an optional audible alarm that sounds when the designated time expires. However, it is important to note that the timer will not automatically terminate the meeting; it is simply a reminder tool for time management.

The new timers are currently available exclusively through the Google Meet web interface on desktop, with Google confirming they are not yet accessible on the mobile applications. A full rollout is underway and is expected to reach all types of Google accounts, including free personal accounts and paid Workspace users, over the next few days.

This small but significant update underscores Google’s commitment to improving the core functionality of its video conferencing platform. For teams looking to shave valuable minutes off their weekly commitments, the new timer is an essential tool for maintaining focus and efficiency in the digital workplace.

Also Read: 

• Generate a Background Image in Google Meet With AI.
• Google Meet Gets a Personal AI Advisor with New 'Ask Gemini' Feature.

Google Photos AI Set to Revolutionize 'Photo to Video' Feature with Custom Prompts.

Google Photos is reportedly preparing a significant upgrade to its generative AI capabilities, specifically for the popular "Photo to video" tool. The new functionality, uncovered in a recent app teardown, suggests users will soon gain highly detailed creative control over video generation.

This move underscores Google's commitment to embedding sophisticated AI features directly into its core consumer applications. The update could transform how users animate their static memories, moving beyond simple preset effects to truly customized digital storytelling.

From Subtle Movements to Creative Commands.

The existing "Photo to video" tool is already powerful, leveraging AI to turn still images into short, dynamic clips. It currently offers two main options for users: "Subtle movements" and the randomized "I'm feeling lucky" effect.

While functional, these choices are creatively restrictive, often requiring multiple tries to get a desired result. The feature recently received an internal boost with the integration of the Veo 3 video generation model.

App Teardown Reveals New Prompt Box

Evidence for the impending update was discovered within an app teardown of Google Photos version 6.48, conducted by Android Authority. This analysis revealed a clear visual element indicating the introduction of a text prompt box.

This prompt box would allow users to describe in detail the action, style, or effect they want applied to their image, unlocking a level of customization previously unavailable. This shift grants users the "creative muscles" to dictate the outcome, making the tool much more versatile.

Expanding Generative AI Access

If released, the prompt-based generation will likely skyrocket the feature's popularity by ensuring greater user satisfaction with the generated results. Users will no longer be reliant on the luck of the draw but on the specificity of their text input.

However, current restrictions on Google’s generative AI tools still apply. Features like "Photo to video" and the "Remix" image editor are currently limited to users in the United States.

Furthermore, generation limits are expected to be in place. While all users will receive a "limited number" of free daily generations, subscribers to the Google AI Pro or Ultra plans will likely benefit from increased usage caps, aligning with Google's broader AI monetization strategy. While not guaranteed for public release, the discovery strongly suggests Google is actively developing this powerful, customizable AI video creation feature.

YouTube Music Begins Testing In-App Lyrics Translation Tool.

YouTube Music is currently rolling out a significant new feature to a limited number of users: native lyrics translation. The long-awaited tool is designed to eliminate language barriers, allowing fans globally to understand the meaning behind songs in languages they do not speak.

For music enthusiasts who enjoy international genres like K-pop or Latin music, this feature offers a major boost to accessibility. Until now, users were limited to viewing lyrics only in the song’s native language when they were available.

The new functionality is accessed directly within the lyrics view of the app. Users have spotted a "Translate" option that, once selected, instantly generates a machine translation of the text, displaying it directly beneath the original lyrics.

Interestingly, the translation’s output language is not manually selectable within the tool itself. Instead, it defaults to the primary language set for the user's system or the specific language configured for the YouTube Music application.

A key detail of this initial test is the restricted access. Reports indicate that the lyrics translation feature is currently only available to users who are subscribed to YouTube Music Premium. This limitation suggests Google may be testing it as a potential perk for paying members.

As this is an early and limited test, the final scope of the feature, including its availability to non-Premium users, remains to be seen. However, its existence signals Google’s commitment to making its music platform more globally inclusive.

Google and Apple Remove 'Red Dot' ICE Alert App Over Policy Violations.

Google has removed the "Red Dot" application, an alert tool designed to track U.S. Immigration and Customs Enforcement (ICE) activity, from its Play Store. The decision follows a similar move by Apple, effectively pulling the app from both major mobile ecosystems.

The app's developers marketed Red Dot as a "safety tool" that would alert users to nearby ICE activity. It reportedly aggregated information from both "reputable sources" and anonymous, user-generated reports of agent movements.

A spokesperson for Google confirmed the removal to Android Authority, stating that the app was in violation of the Play Store policy. The company specifically cited its rule against applications that carry a "high risk of abuse."

Furthermore, Google noted that apps featuring user-generated content, as Red Dot did with its anonymous reporting system, are subject to stringent moderation requirements. The platform concluded that Red Dot failed to meet these necessary content control standards.

The removal comes amid heightened public interest and scrutiny regarding the aggressive tactics of federal immigration enforcement. However, Google clarified that the Department of Justice had not contacted the company regarding the app prior to its takedown.

This incident is not isolated; the takedown mirrors Apple's earlier removal of the iOS-exclusive app "ICEBlock," which performed a similar function. This signals a unified stance by the major mobile platforms on applications involving law enforcement agencies' location-tracking.

Discord Customer Service Provider Hacked, Exposing User Data and Government IDs.

Discord, the popular voice, video, and text communication platform, has recently disclosed a security incident stemming from a breach at one of its third-party customer service providers. The company confirmed that an unauthorized party gained access to limited user information, impacting individuals who had recently interacted with its Customer Support or Trust & Safety teams.

The company emphasized that the breach did not compromise Discord’s core systems or platform directly. The attackers targeted the external vendor responsible for handling support tickets, with the ultimate goal of extorting a financial ransom from Discord.

Data exposed in the incident includes essential contact information provided during support requests. Compromised data may consist of user names, Discord usernames, email addresses, IP addresses, and the full content of messages exchanged with customer service agents.

Most alarmingly, the breach exposed a "small number" of government-issued photo identification documents, such as passports and driver's licenses. These sensitive files belonged to users who had submitted them to appeal age determinations on the platform. Limited billing information, specifically the payment type, the last four digits of a credit card, and purchase history, was also compromised for some users.

Crucially, Discord confirmed that user passwords, full credit card numbers, CCV codes, and private in-app messages and activity outside of customer support threads were not affected by the breach. This containment ensures that core account credentials remain secured.

Upon discovering the compromise, Discord immediately revoked the third-party provider's access to its ticketing system. The company has since launched a full internal investigation, engaged a leading computer forensics firm, and notified law enforcement to pursue the unauthorized party.

Discord is currently in the process of notifying all affected users via email. Impacted users should look for an official message from noreply@discord.com that will specify exactly what data, including whether their government ID, may have been accessed. Users are strongly advised to remain vigilant against potential phishing attempts.