ML Kit

AppWizard
July 31, 2026
Bitsight's investigation revealed that inexpensive Android TV boxes are being shipped with applications that can change their hardware identity, allowing them to impersonate popular smartphone brands like Samsung and Huawei. This operation, named Fuyao, is linked to Zhejiang Fengwo IoT Technology Co., Ltd., a Chinese company established in 2019. The H96MAXV11 model was frequently reported among the affected devices. In one day, the operation received 65,957 reports from about 38,000 unique MAC addresses, with many devices misidentified as phones due to spoofed identifiers. Fengwo has also promoted over 120,000 "AI digital humans," though details on this marketing term are vague. The command-and-control server for Fuyao sends phone profiles to devices, masking their actual hardware specifications. The operation uses machine vision technology and a YOLOv8s object-detection model to identify advertisements. Bitsight documented 40 fraud tasks, 21 unique campaigns, and 166 modules across four devices. The operation's payout structure involves 144 operator-owned domains, with an estimated gross return of .25 per device daily, potentially leading to annual revenues in the millions. Attribution to Fengwo is supported by shared TLS certificate data and public patent records, although the patents do not directly address advertising. There is uncertainty about how the fraudulent apps were installed and at what point in the supply chain they were introduced. Device owners are advised to verify Play Protect certification and disconnect suspicious devices.
AppWizard
June 4, 2025
Google has updated the Gemini Nano APIs through the ML Kit, introducing new GenAI APIs that allow developers to perform on-device inference for generative AI capabilities in Android applications. These APIs enhance user privacy by executing all operations locally, eliminating cloud processing costs. Features include summarization of articles up to 3,000 words, proofreading for grammatical accuracy, rewriting text in various tones, and generating image descriptions. The GenAI APIs are optimized for Android devices using MediaTek Dimensity, Qualcomm Snapdragon, and Google Tensor platforms, including the Pixel 9 series and Samsung Galaxy S25. They operate without internet connectivity, ensuring rapid response times and improved performance for tasks like summarization and proofreading. The local processing also enhances privacy and security, protecting sensitive user information with safety protocols.
AppWizard
May 17, 2025
Google announced on May 14, 2025, the expansion of its ML Kit offerings with new GenAI APIs powered by the Gemini Nano model, aimed at providing Android developers with access to advanced AI features while ensuring user privacy through on-device processing. The ML Kit GenAI APIs include four key capabilities: summarization, proofreading, rewriting, and image description. These APIs process data locally, keeping user data on the device, ensuring consistent functionality without internet connectivity, and avoiding recurring server costs. This development enhances the accessibility of sophisticated AI capabilities for Android developers and aligns with consumer privacy concerns. The architecture allows for future expansion of capabilities as Gemini models evolve.
AppWizard
May 16, 2025
Google is preparing to unveil new mobile AI tools at the upcoming I/O event, including a suite of APIs that will enable developers to use Gemini Nano for on-device AI applications. The ML Kit SDK will be updated to support on-device generative AI functionalities through Gemini Nano, which integrates with existing models and offers predefined features for ease of implementation. The ML Kit’s GenAI APIs will allow applications to perform tasks such as summarization, proofreading, rewriting, and image description without cloud data transmission. However, Gemini Nano's capabilities are limited compared to cloud-based options, with summaries restricted to three bullet points and image descriptions available only in English. The standard version, Gemini Nano XS, requires about 100MB of storage, while the smaller Gemini Nano XXS is text-only and occupies a quarter of that size. ML Kit is compatible with devices beyond Google's Pixel lineup, including the OnePlus 13, Samsung Galaxy S25, and Xiaomi 15, providing developers opportunities to enhance applications with generative AI features.
Search