February 8 Tech news Roundup: ChatGPT Search and Deep Research are ready to rock the internet, Google Gemini 2.0 is here, Malicious mobile apps can use OCR to steal data

ChatGPT Search no longer requires an account

This week was pretty big for OpenAI, the company has made ChatGPT Search available for all users. The service was previously locked behind a requirement: an OpenAI account, now you no longer need to sign up for an account to access the AI-powered search engine. Users who wish to use ChatGPT to search the web can visit ChatGPT.com and enter their queries in the chat box, and click on the Search button that appears below it. ChatGPT Search will present the result similar to how its chatbot replies to your questions. The results are displayed as rich snippets with a brief text summary, and an image from the link it was sourced from. A key difference between ChatGPT and ChatGPT Search is that the latter pulls data from the internet in real-time, as opposed to being trained on a specific data set.

While this is a nice way to test the capabilities of the search engine, it does have some limitations, such as not being able to view your search history. The AI-driven search engine looks set to compete with the likes of Google Search, Bing, DuckDuckGo, and Brave Search.

ChatGPT on WhatsApp now supports voice messages and images. Users can sign in to their account to access premium features available for ChatGPT Plus and Pro.

OpenAI launches Deep Research for ChatGPT

OpenAI has unleashed Deep Research for ChatGPT users. It is a cutting-edge technology that is capable of offering complex, in-depth research across various topics. The AI uses the latest o3 model, and has been trained using multi-layered neural networks to analyze patterns, make decisions by reasoning, and perform tasks like a human.

Users will need to select “deep research” in the ChatGPT composer, to perform an in-depth search. However, it does not answer queries instantly like the regular ChatGPT. Deep Research is significantly slower, and could take up to 30 minutes to produce an answer. Deep Research is not free at the moment, it is currently available for ChatGPT Pro subscribers. There is a limit of 100 queries per month. OpenAI has acknowledged that the feature isn’t flawless, it can result in mistakes, or misinterpret sources, and the report formatting is not perfect. The company is working on improving the feature by offering support for data visualizations, embedded images, and more advanced data sources.

Google’s Gemini 2.0 is here

Google has released Gemini 2.0 Flash for all users. Gemini is now more efficient than ever and also offers better performance.  The new AI-model is available via the Gemini app on mobile devices and on the web.

Gemini 2.0 Pro Experimental is now available in the Gemini app for advanced users, as well as Google AI Studio, Vertex AI. The experimental model can handle complex prompts and coding, thanks to its 2-million-token context window. It is capable of deep analysis of large data sets, understanding and reasoning. Google has also debuted Gemini 2.0 Flash-Lite, which is its most cost-efficient AI model. It has a 1-million-token context window, and supports multimodal capabilities, and can handle tasks like generating captions for thousands of images. Google is working to add more multimodal functionalities to Gemini, such as image generation and text-to-speech capabilities.

Gemini does have its fair share of problems, as Google admitted that hackers are using the AI-powered tool for cybercrime activities such as researching vulnerabilities, drafting phishing campaigns, and gathering intelligence on defense organizations.

Google Search to get AI-powered assistant features soon

Google Search will soon get AI-powered features. The first of the planned features includes an option to ask a follow-up question, allowing users to interact with the service similar to how one might use an AI assistant. Google’s upcoming features include Project Mariner, which is an AI agent that can navigate a browser, click buttons and fill out forms on its own.

OpenAI isn’t the only one that has an in-depth research tool, Google has its own called “Gemini Deep Research”, which is designed to assist users in their academic or professional needs to generate detailed reports that contain source links, key findings, and can export content to Google Docs.

YouTube is testing 4x playback speeds for some reason

In what appears to be a rather bizarre move, YouTube is testing an option to play videos at 4x playback speeds. Up until now, YouTube has allowed users to play videos at a maximum speed of 2x the normal time. Google is also testing a feature called “Jump Ahead” to assist viewers get to their preferred content quickly. YouTube has added an audio quality enhancement with support for 256kbps bitrate in music videos. iOS users can now watch YouTube shorts in picture-in-picture mode, while they multitask with other apps. Google has also enabled Smart Downloads to automate video downloads for Shorts, to allow users to watch the videos when they are offline.

YouTube 4x playback speeds are currently available for Premium subscribers. It is unclear why this feature was designed, and only time will be able to tell us if it is popular among users. Experimental features like the ones described above are tested temporarily, before they are either removed, or released for all users.

SparkCat mobile malware used infected apps and OCR to steal data

Security researchers at Kaspersky have discovered that malicious apps on the Google Play Store, Apple App Store, and third-party channels are using a new method to steal data. The new malware-strain called SparkCat, which had been active since April 2024, were written in the Rust programming language and had a malicious SDK. When a user installed an infected app on their phone, it would silently install an OCR (optical character recognition) plugin on the device. The malware would use it to scan images on the user’s phone to steal recover codes for Cryptocurrency. These malware were seemingly designed to target Android and iPhone users in Europe and Asia, and the apps were operating in several countries.

Google had blocked 2.3 Million risky apps in 2024. Apple has confirmed that it pulled 11 apps from the iOS App Store including ComeCome, WeTink, and AnyGPT, to protect iPhone users from the SparkCat malware attacks.