Habr•July 19, 2026•🇷🇺Translated from Russian

Mimolet Dating App Review Highlights Privacy Protections, AI Moderation, and UX Trade-Offs in Detailed Analysis

Mimolet presents itself as a standard dating application featuring a vertical feed of user profiles. However, a closer examination after registration reveals a broader set of functionalities that extend well beyond simple swiping, including detailed profiles, full-featured chats, public groups, integrated voice and video calls, AI tools, and a dedicated moderation system.

Registration, Profiles, and Vertical Feed

Registration in Mimolet requires more steps than many competing dating apps. In addition to name, city, photos, and dating goals, users must specify height and select between three and eight interests from a list of approximately ninety options covering games, music, films, fashion, and walks. Optional fields include favorite artists, attitudes toward smoking and alcohol, subculture affiliation, and willingness to meet in person, which helps set clear expectations from the start.

The vertical feed displays profiles in a scrolling format similar to short videos, with each card occupying most of the screen. Key information such as city, height, dating goals, and interests appears directly beneath the main photo, eliminating the need to open a separate profile view before deciding to engage.

Filters, Appearance-Based Search, and Communication Tools

Most filters, including those for city, height, interests, dating goals, and subculture, remain available without a subscription. A weight filter exists but stands out as potentially unnecessary and intrusive. An appearance-based search feature allows users to upload a reference photo and receive up to ten matching profiles with similarity percentages, though the app could better clarify what happens to the uploaded image after the search completes.

Chats function like a full messenger with replies, reactions, editing, deletion by either party, voice messages, video circles, stickers, archiving, and pinned conversations. Three AI tools assist with message suggestions, grammar correction, and voice-to-text transcription with punctuation. Built-in calls use WebRTC and temporary identifiers so users can speak without exchanging phone numbers.

Groups, Events, and Social Features

Public groups organized by cities, interests, and subcultures enable conversations to begin around shared topics rather than requiring an initial match. Users are automatically added to relevant groups during profile setup but can leave permanently. Events listing allows users to mark attendance and join associated chats, while a voice roulette connects random participants and a connections feature displays confirmed friends and partners with mutual consent required.

AI Characters, Moderation, and Data Handling

Several AI characters, clearly marked with an AI badge, provide app support, conversational practice, or tarot readings. Complaints and blocking remain free, with photos reviewed by AI models before publication. Images are processed on the server to determine true format, resized, and stripped of EXIF metadata containing device details, timestamps, and location data. The main application, API, and database reside in a Russian server environment, while media files are stored in a Russian S3-compatible cloud.

Privacy Questions and Monetization

Two notable privacy gaps remain: the privacy policy lacks specific retention periods for photos, backups, moderation materials, and AI conversation history, and blocked users have no dedicated appeals form with case numbers or status tracking. Three subscription tiers—Plus, Premium, and Ultra—increase reaction limits, boost visibility, and unlock additional AI capabilities while core features stay free. Monetization also includes blurred incoming likes that require payment to view and an optional ten-point photo rating system.

According to the development team, the service has around 200,000 registrations and roughly 15,000 daily active users, with retention rates of 44.96% on day three, 34.69% on day seven, and 19.11% on day thirty.

Related articles

Habr•Privacy & Surveillance

Pocket Android: Open-Source App Uses Work Profiles to Isolate Apps Behind Dedicated Proxies

A developer has released Pocket Android, a free open-source application that leverages Android's built-in Work Profile feature to create an isolated environment for selected apps. The tool routes traffic from cloned applications exclusively through user-defined proxies while keeping banking, delivery, and government services in the main profile on direct connections. It addresses common failures of per-app exclusion lists by enforcing strict lockdown policies that prevent leaks after reboots or proxy drops. The project builds on the Shelter codebase and integrates Xray for protocol support including VLESS, REALITY, and Hysteria 2. Battery impact is minimized through event-driven status checks rather than constant timers. The approach provides practical privacy separation for services sensitive to IP changes such as Claude and ChatGPT without requiring manual proxy switching throughout the day.

AntiMalware•Privacy & Surveillance

Telegram Beta for macOS Introduces Hidden Chats Feature Protected by Secret Passphrase

The beta version of Telegram 13.2.0 for macOS now includes a Hidden Chats function that lets users move selected chats and channels out of the main list into a separate section accessible only via a secret passphrase. Users configure the feature in the privacy settings by entering a passphrase of at least eight Latin characters twice. Multiple passphrases can be created, each unlocking its own set of hidden conversations. The hidden section automatically closes after screen lock, sleep mode, or five minutes of inactivity, and calls from hidden chats are silenced. The passphrase is not stored anywhere on the device, so forgotten credentials require logging out and back in to regain access on that computer. The feature is device-specific and does not sync across platforms, meaning hidden chats remain visible in the main list on smartphones and other clients.

Habr•Privacy & Surveillance

Windows 11 Telemetry Secretly Tracks AI Coding Agents and Developer Tools via MissionControlAggregator

Reverse engineering of Windows 11 binaries revealed a hardcoded list of coding agents and developer tools inside MissionControlAggregator.dll. The component monitors launches of applications such as codex.exe, claude.exe, copilot.exe, cursor.exe, python.exe, gcc.exe and code.exe through existing ETW events from KernelProcess and Win32kTraceLogging. Detected usage is stored as bit flags in SubCohorts reports and sent to Microsoft when optional diagnostic data is enabled, which remains the default on US installations. Researchers confirmed the mechanism by renaming cmd.exe to codex.exe and observing the corresponding bit set in Segments.SubCohorts events queued by DiagTrack. The feature CLICodingAgentUsage (ID 62580080) categorizes both CLI and desktop agents without collecting prompts or source code. Data collection occurs on standard Windows 11 ARM64 builds and persists in AggregatorStorage even without user interaction with the listed tools.

BoletimSec•Privacy & Surveillance

Meta Fined €805 Million by Italian Data Protection Authority for Unauthorized Access to Android Photo Gallery

The Italian data protection authority Garante has imposed an €805 million fine on Meta for improper access to users' photo galleries via the Facebook Android app. The feature, active by default since 2023, allowed continuous reading of images and metadata without explicit, separate consent from approximately 20 million Italian Android users. The authority ruled that the single broad permission request violated Article 122 of the Italian Privacy Code, which implements the EU ePrivacy Directive and treats device storage as part of private life. Meta must now implement separate consent mechanisms, provide clear explanations, and keep the feature disabled by default within 60 days or face further restrictions. The company plans to appeal the decision, describing the penalty as disproportionate, while the investigation originated from a consumer complaint in Bolzano and included technical analysis by the regulator. The fine accounts for the scale of affected users, the ongoing nature of the violation, prior enforcement actions, and limited cooperation.