Latest news with #MLKit
Android Authority
16-05-2025
- Business
- Android Authority
Google is about to unleash Gemini Nano's power for third-party Android apps
Edgar Cervantes / Android Authority TL;DR Google is expanding access to Gemini Nano, its on-device AI model, through new ML Kit GenAI APIs. These new APIs, likely to be announced at I/O 2025, will enable developers to easily implement features like text summarization, proofreading, rewriting, and image description generation in their apps. Unlike the experimental AI Edge SDK, ML Kit's GenAI APIs will be in beta, support image input, and be available on a wider range of Android devices beyond the Pixel 9 series. Generative AI technology is changing how we communicate and create content online. Many people ask AI chatbots like Google Gemini to perform tasks such as summarizing an article, proofreading an email, or rewriting a message. However, some people are wary of using these AI chatbots, especially when these tasks involve highly personal or sensitive information. To address these privacy concerns, Google offers Gemini Nano, a smaller, more optimized version of its AI model that runs directly on the device instead of on a cloud server. While access to Gemini Nano has so far been limited to a single device line and text-only input, Google will soon significantly expand its availability and introduce image input support. Late last month, Google published the session list for I/O 2025, which includes a session titled, 'Gemini Nano on Android: Building with on-device gen AI.' The session's description states it will 'introduce a new set of generative AI APIs that harness the power of Gemini Nano. These new APIs make it easy to implement use cases to summarize, proofread, and rewrite text, as well as to generate image descriptions.' In October, Google opened up experimental access to Gemini Nano via the AI Edge SDK, allowing third-party developers to experiment with text-to-text prompts on the Pixel 9 series. The AI Edge SDK enables text-based features like rephrasing, smart replies, proofreading, and summarization, but it notably does not include support for generating image descriptions, a feature Google highlighted for the upcoming I/O session. Thus, it's likely that the 'new set of generative AI APIs' mentioned in the session's description refers to either something entirely different from the AI Edge SDK or a newer version of it. Fortunately, we don't have to wait until next week to find out. Earlier this week, Google quietly published documentation on ML Kit's new GenAI APIs. ML Kit is an SDK that allows developers to leverage machine learning capabilities in their apps without needing to understand how the underlying models work. The new GenAI APIs allow developers to 'harness the power of Gemini Nano to deliver out-of-the-box performance for common tasks through a simplified, high-level interface.' Like the AI Edge SDK, it's 'built on AICore,' enabling 'on-device execution of AI foundation models like Gemini Nano, enhancing app functionality and user privacy by keeping data processing local. Mishaal Rahman / Android Authority In other words, ML Kit's GenAI APIs make it simple for developers to use Gemini Nano for various features in their apps privately and with high performance. These features currently include summarizing, proofreading, or rewriting text, as well as generating image descriptions. All four of these features match what's mentioned in the I/O session's description, suggesting that Google intends to formally announce ML Kit's GenAI APIs next week. Here's a summary of all the features offered by ML Kit's GenAI APIs: Summarization : Summarize articles or chat conversations as a bulleted list. Generates up to three bullet points Languages: English, Japanese, and Korean : Summarize articles or chat conversations as a bulleted list. Proofreading : Polish short content by refining grammar and fixing spelling errors. Languages: English, Japanese, German, French, Italian, Spanish, and Korean : Polish short content by refining grammar and fixing spelling errors. Rewrite : Rewrite short chat messages in different tones or styles. Styles: Elaborate, Emojify, Shorten, Friendly, Professional, Rephrase Languages: English, Japanese, German, French, Italian, Spanish, and Korean : Rewrite short chat messages in different tones or styles. Image description : Generate a short description of a given image. Languages: English : Generate a short description of a given image. Compared to the existing AI Edge SDK, ML Kit's GenAI APIs will be offered in 'beta' instead of 'experimental access.' This 'beta' designation could mean Google will allow apps to use the new GenAI APIs in production. Currently, developers cannot release apps using the AI Edge SDK, meaning no third-party apps can leverage Gemini Nano at this time. Another difference is that the AI Edge SDK is limited to text input, whereas ML Kit's GenAI APIs support images. This image support enables the image description feature, allowing apps to generate short descriptions of any given image. The biggest difference between the current version of the AI Edge SDK and ML Kit's GenAI APIs, though, is device support. While the AI Edge SDK only supports the Google Pixel 9 series, ML Kit's GenAI APIs can be used on any Android phone that supports the multimodal Gemini Nano model. This includes devices like the HONOR Magic 7, Motorola Razr 60 Ultra, OnePlus 13, Samsung Galaxy S25, Xiaomi 15, and more. Developers who are interested in trying out Gemini Nano in their apps can get started by reading the public documentation for the ML Kit GenAI APIs. Got a tip? Talk to us! Email our staff at Email our staff at news@ . You can stay anonymous or get credit for the info, it's your choice.
Yahoo
06-02-2025
- Yahoo
Screenshot-reading malware cracks iPhone security for the first time
In the realm of smartphones, Apple's ecosystem is deemed to be the safer one. Independent analysis by security experts has also proved that point repeatedly over the years. But Apple's guardrails are not impenetrable. On the contrary, it seems bad actors have managed yet another worrying breakthrough. As per an analysis by Kaspersky, malware with Optical Character Recognition (OCR) capabilities has been spotted on the App Store for the first time. Instead of stealing files stored on a phone, the malware scanned screenshots stored locally, analyzed the text content, and relayed the necessary information to servers. The malware-seeding operation, codenamed 'SparkCat,' targeted apps seeded from official repositories — Google's Play Store and Apple's App Store — and third-party sources. The infected apps amassed roughly a quarter million downloads across both platforms. Interestingly, the malware piggybacked atop Google's ML Kit library, a toolkit that lets developers deploy machine learning capabilities for quick and offline data processing in apps. This ML Kit system is what ultimately allowed the Google OCR model to scan photos stored on an iPhone and recognize the text containing sensitive information. But it seems the malware was not just capable of stealing crypto-related recovery codes. 'It must be noted that the malware is flexible enough to steal not just these phrases but also other sensitive data from the gallery, such as messages or passwords that might have been captured in screenshots,' says Kaspersky's report. Among the targeted iPhone apps was ComeCome, which appears to be a Chinese food delivery app on the surface, but came loaded with a screenshot-reading malware. 'This is the first known case of an app infected with OCR spyware being found in Apple's official app marketplace,' notes Kaspersky's analysis. It is, however, unclear whether the developers of these problematic apps were engaged in embedding the malware, or if it was a supply chain attack. Irrespective of the origin, the whole pipeline was quite inconspicuous as the apps seemed legitimate and catered to tasks such as messaging, AI learning, or food delivery. Notably, the cross-platform malware was also capable of obfuscating its presence, which made it harder to detect. The primary objective of this campaign was extracting crypto wallet recovery phrases, which can allow a bad actor to take over a person's crypto wallet and get away with their assets. The target zones appear to be Europe and Asia, but some of the hotlisted apps appear to be operating in Africa and other regions, as well.
