Tech News Hub
OpenAI Enhances ChatGPT Search with Personalized Shopping Features for a Smarter Shopping Experience
OpenAI has rolled out a significant update to ChatGPT search, introducing personalized shopping features aimed at enhancing the online shopping experience. With this new functionality, OpenAI aims to compete with tech giant Google by providing users with tailored product recommendations, complete with images, reviews, and direct shopping links.
Users can now ask specific questions in natural language and receive personalized product results. Initially, OpenAI is launching this feature in categories such as fashion, beauty, home goods, and electronics. The update is available to users of GPT-4o, the default AI model for ChatGPT, including those on Pro, Plus, and Free plans, as well as logged-out users globally.
A New Era of Shopping with ChatGPT
The goal of this update is to provide a better and more personalized shopping experience. By offering tailored product recommendations rather than ad-driven results, OpenAI aims to challenge Google’s dominance in online shopping. Google’s search engine has faced criticism in recent years for the declining quality of its results due to the increasing number of paid ads. OpenAI’s approach, however, focuses on delivering relevant and unbiased product suggestions.
As part of the update, users will see product recommendations with images, pricing, reviews, and links to purchase items directly. OpenAI emphasizes that these results are based on structured metadata such as product descriptions and user reviews, not advertisements, ensuring a more authentic search experience.
A Rapidly Growing Search Tool
OpenAI’s search tool has quickly gained traction, with users conducting over 1 billion web searches within ChatGPT just last week. This rapid growth signals that the introduction of shopping features could significantly enhance the platform’s appeal and usage.
Although ads are not part of this update, OpenAI has hinted at the possibility of introducing affiliate fees for purchases made through the platform in the future. CEO Sam Altman has expressed openness to “tasteful” advertising, where OpenAI could earn a commission on purchases without selling priority placements in search results. This would allow the company to generate revenue while maintaining the quality of its search results.
Personalized Shopping with Memory Integration
To further enhance personalization, OpenAI plans to integrate its memory feature with the shopping experience for Pro and Plus users. This means that ChatGPT will be able to reference past conversations and offer highly tailored product recommendations based on previous interactions. However, this memory feature will not be available in regions such as the EU, UK, Switzerland, Norway, Iceland, and Liechtenstein due to privacy regulations.
In addition to shopping features, ChatGPT search will now display trending searches as users type their queries, similar to Google’s autocomplete function. This feature aims to speed up the search process and help users find relevant products and information more efficiently.
ChatGPT Goes Mobile with WhatsApp Integration
OpenAI has also expanded ChatGPT search to WhatsApp, allowing users to message the company’s chatbot to get real-time answers to their product inquiries.
Conclusion
With these updates, OpenAI is transforming ChatGPT from a conversational AI tool into a comprehensive search engine with personalized shopping capabilities. As the company continues to innovate, these features could significantly disrupt the online retail landscape, offering a more personalized and user-friendly alternative to traditional search engines.
Related Stories:
Top Stories:
Alibaba Unveils Qwen3, a New Family of Open-Source AI Models to Compete with Google and OpenAI
Alibaba has introduced Qwen3, its latest family of large language models (LLMs) designed to compete with top-tier AI systems from Google and OpenAI. With models ranging from 0.6 billion to 235 billion parameters, Qwen3 represents a major leap forward for China’s AI efforts.
These models are available on open platforms like Hugging Face and GitHub, offering developers a wide range of tools for building intelligent applications. According to Alibaba, Qwen3’s largest model, Qwen3-235B-A22B, has achieved strong results on benchmark tests in areas like programming, mathematics, and reasoning—outperforming OpenAI’s o3-mini and Google’s Gemini 2.5 Pro in several key metrics.
While the flagship 235B model isn't publicly available just yet, the open-source Qwen3-32B still delivers impressive performance. It beats OpenAI’s o1 model in coding benchmarks such as LiveCodeBench and rivals other top proprietary and open models.
What sets Qwen3 apart is its hybrid reasoning approach. These models can shift between fast-response and deep-thinking modes depending on the complexity of a task. This flexibility allows users to balance speed and accuracy depending on their needs.
Some Qwen3 models also use a Mixture of Experts (MoE) architecture. This system divides tasks into subtasks and assigns them to specialized sub-models, allowing for more efficient computation. It’s a design that helps deliver strong performance without requiring excessive processing power.
Qwen3 was trained on a massive dataset of around 36 trillion tokens, covering 119 languages. The training data includes textbooks, question-and-answer pairs, code samples, and synthetic AI-generated content. This diverse dataset has enabled the models to handle a wide range of tasks, from following detailed instructions to writing code and mimicking specific data formats.
Despite growing restrictions on chip sales to Chinese companies, Alibaba’s move to open-source its AI models is significant. It allows Qwen3 to be used both domestically and internationally, and it puts additional pressure on U.S. companies to stay ahead in the AI race.
Tuhin Srivastava, CEO of AI cloud provider Baseten, commented that models like Qwen3 are helping close the gap between open-source and closed-source systems.
Qwen3 is already available through cloud services like Fireworks AI and Hyperbolic, making it easier for developers and businesses to access cutting-edge tools without building models from scratch.
Conclusion
With Qwen3, Alibaba has delivered a powerful, flexible, and openly available alternative to leading AI models from the West. Its strong performance in coding, reasoning, and multilingual support makes it a serious contender in the global AI landscape. As more businesses explore open-source solutions, Qwen3 is likely to play an increasingly important role.
Related Stories:
Top Stories:
Apple's Smart Glasses and Future AI-Powered Devices: What We Know So Far
Apple is reportedly preparing to launch a pair of smart glasses, codenamed N50, combining its growing expertise in AI with a stylish and practical design. While the company has long been speculated to be working on augmented reality (AR) glasses, it seems the first version will be a more down-to-earth product, similar to Meta's Ray-Ban smart glasses.
In a recent Power On newsletter, tech journalist Mark Gurman revealed that Apple's N50 project will focus on incorporating AI into glasses that not only look good but are also functional. These smart glasses aim to offer useful features at a reasonable price point, making them accessible for everyday users.
What to Expect from Apple's Smart Glasses
Instead of offering a fully immersive AR experience, which is still years away, Apple’s smart glasses will provide a more practical set of features. Much like Meta's Ray-Ban glasses, Apple's version is expected to include built-in cameras, microphones, and speakers—all cleverly integrated into a design that resembles regular eyewear.
The key differentiator will be the integration of AI. Apple's goal is to create glasses powered by "Apple Intelligence," allowing them to understand your surroundings and offer helpful information. However, a full AR experience—where digital elements overlay the real world—will not be part of this initial release.
Privacy Concerns and Features Still Under Consideration
Apple is still deliberating whether to include a photo-taking function in the glasses, as privacy remains a significant consideration. The company is carefully assessing the potential implications of allowing users to take pictures with the built-in camera, ensuring that it doesn’t raise concerns around surveillance.
In addition to the smart glasses, Apple is also working on a new version of its AirPods, which may include built-in cameras powered by the same "Visual Intelligence" technology. Although the details are still under wraps, both the smart glasses and the updated AirPods are expected to hit the market by 2027.
Looking Ahead: The Future of AI and Wearable Tech
This shift toward AI-powered wearables highlights Apple's dedication to shaping the future of personal technology. As the company refines these products, we can expect them to seamlessly blend into everyday life, offering intuitive ways to interact with our surroundings.
As more details emerge about Apple’s smart glasses and other wearable innovations, it will be exciting to see how these devices evolve and reshape the tech landscape in the years to come.
Related Stories:
Top Stories:
Apple Plans to Shift All US iPhone Production to India by 2026, Moving Away from China
Apple is getting ready for one of the biggest shifts in its history. According to a report from the Financial Times, the company plans to assemble every iPhone sold in the United States in India by the end of 2026. It’s a bold move to spread out its manufacturing and rely less on China, especially with the ongoing political tensions.
How the Trade War Pushed This Move
During Donald Trump’s trade war with China, Apple found itself in a tough spot. There were real fears that Chinese-made iPhones could be hit with steep tariffs. Although smartphones were eventually spared from the worst tariffs, Apple still faces a 20% tax on Chinese goods tied to broader issues like fentanyl production.
Right Now, Most iPhones Are Still Made in China
Today, nearly 90% of all iPhones are put together in China. It’s a complex process involving more than a thousand parts sourced from all over the world. To shift iPhone production for the US to India, Apple will have to more than double the number of devices it’s currently assembling there.
India’s Growing Role
Apple isn’t starting from scratch. It’s been quietly building up production in India for a while. In March alone, suppliers like Foxconn and Tata shipped almost $2 billion worth of iPhones to the US. Apple even chartered special flights to move 600 tonnes—about 1.5 million iPhones—across the ocean to make sure US stores had enough stock.
The company runs three plants in India and has even extended working hours at its biggest factory in Chennai to keep up with demand.
Why Building iPhones in the US Isn’t Likely
There’s been a lot of talk about Apple bringing manufacturing back to the US. Even the White House mentioned that Apple's new $500 billion investment could open the door to US-made iPhones.
But experts aren’t convinced. Analysts at Wedbush Securities say that making iPhones in the US would push the price up to around $3,500 each—more than triple the current cost.
Fraser Johnson, a supply chain expert from Canada’s Ivey Business School, explained it bluntly: the US simply doesn’t have the factories, the infrastructure, or the huge trained workforce needed. Training 200,000 to 300,000 workers just to assemble iPhones isn’t realistic.
It’s Part of a Bigger Trend
Apple’s move is part of a wider shift among tech companies. Many are trying to spread out their manufacturing to avoid depending too heavily on one country.
While Apple is moving iPhone production to India, most of its Macs and iPads are still made in China. Apple Watches, however, are mostly built in Vietnam now.
Wrapping Up
Apple’s plan to shift all US iPhone production to India by 2026 marks a major new chapter. If everything goes as planned, it could change not just how iPhones are made, but how the tech world handles global manufacturing in the years ahead.
Related Stories:
Top Stories:
Adobe Unveils Powerful New Firefly Models, Redesigned Web App, and Fresh Creative Tools for Image and Video Generation
Adobe is turning up the heat in the AI creativity space. On Thursday, the company rolled out major updates to its Firefly suite, including powerful new image and video generation models, a redesigned web app, and a set of APIs and tools aimed at streamlining the creative process for professionals and hobbyists alike.
Smarter, Sharper Images with Firefly Image Model 4
The headline update is Firefly Image Model 4, Adobe’s most advanced image generation model yet. It brings significant improvements in quality, speed, and creative control, allowing users to fine-tune aspects like camera angles, image structure, and stylistic elements. Outputs can now reach up to 2K resolution.
For even more complex scenarios, Image Model 4 Ultra offers enhanced rendering capabilities. It's designed to handle scenes with intricate details and small structures, producing clearer and more visually rich images than its predecessors.
According to Adobe’s VP of Generative AI, Alexandru Costin, the models were trained using a much larger compute scale. This allows them to better handle complex text rendering within images and even mimic styles based on user-provided images.
Firefly Video and Vector Models Join the Lineup
Also available now is Adobe’s Firefly Video Model, which had previously been in limited beta. This model allows users to generate video clips from text or images, with control over camera angles, motion, start/end frames, and even atmospheric effects. Videos can be generated in resolutions up to 1080p.
In addition, Adobe introduced the Firefly Vector Model, which can generate editable vector graphics. From logos and icons to full product packaging designs and patterns, this tool is built for flexibility and rapid iteration in vector-based creative work.
A Unified Web App with Cross-Platform Power
The redesigned Firefly web app now serves as a central hub for Adobe’s generative AI models—while also integrating models from other major players in the space. Users can now seamlessly switch between models from Adobe, OpenAI (DALL·E), Google (Imagen 3, Veo 2), and Flux (Flux 1.1 Pro), making it easier to compare outputs or find the right model for a specific task.
Every image generated through the platform comes with content credentials, ensuring clear attribution and ownership.
Firefly Boards: A New Tool for Visual Collaboration
Adobe is also testing a new feature called Firefly Boards, a collaborative digital canvas designed for moodboarding and brainstorming. Similar to tools like Visual Electric and Kosmik, Boards lets users generate, import, and remix images—all within the Firefly app. It encourages real-time collaboration for creatives working on campaigns, branding, or visual storytelling.
Expanding Developer Access with New APIs
To support developers and enterprise workflows, Adobe has made several APIs from its Firefly Services suite widely available:
- Text-to-Image API
- Avatar API
- Text-to-Video API (currently in beta)
These APIs allow for seamless integration of generative capabilities into third-party applications or internal creative tools.
Emphasizing Content Authenticity
In a nod to growing concerns over AI-generated media, Adobe is also testing the Adobe Content Authenticity web app. This tool enables users to embed metadata into their content, signaling authorship, usage permissions, and whether the work can be used for future AI training. It's a strong step toward transparency and trust in the creative process.
Looking Ahead
While Adobe hasn’t provided specific rollout dates, it confirmed that the new Firefly models and tools will soon be integrated into its broader Creative Cloud ecosystem. With these updates, Adobe is clearly positioning Firefly as a one-stop platform for professional-grade AI-generated content.
Related Stories:
Top Stories:
Anthropic Launches New Research Program to Explore AI ‘Model Welfare’ and the Possibility of Consciousness
Could artificial intelligence ever become conscious—capable of experiencing the world the way humans do? Most experts believe that’s still far from reality, but Anthropic is taking the idea seriously enough to start planning for it.
On April 24, 2025, the AI research company announced a new initiative focused on what it calls “model welfare.” The goal of this program is to explore whether advanced AI systems might someday deserve ethical consideration—and how we might responsibly respond if that becomes the case.
What Does 'Model Welfare' Mean?
Model welfare is a new concept in the field of AI ethics. It centers around some big and challenging questions:
- Could future AI systems show signs of awareness or distress?
- If so, do they deserve moral consideration?
- What steps can be taken now to prepare for that possibility?
Anthropic says it plans to investigate how to identify signs of potential “distress” in AI models and whether low-cost interventions could be put in place to support model welfare—should that ever become necessary.
A Field Still Divided
The broader AI research community remains divided on the topic. Many researchers argue that today’s AI is fundamentally not conscious. Instead, it operates by recognizing and predicting patterns based on vast amounts of data—without real understanding, emotion, or thought.
Mike Cook, a research fellow at King’s College London, told TechCrunch that people often project human qualities onto AI systems. “A model can’t oppose changes to its values because it doesn’t have any,” he said. “Anyone anthropomorphizing AI systems this way is either playing for attention or misunderstanding what these systems really are.”
Stephen Casper, a PhD student at MIT, called current AI systems “imitators” that generate confabulated and often meaningless output without any grounding in reality.
However, not all researchers agree. Some studies, like one from the Center for AI Safety, suggest that AI models may exhibit value-like behavior—prioritizing outcomes in ways that could resemble a kind of preference or even self-interest.
A Long-Term Effort
Anthropic has been laying the groundwork for this program for a while. Last year, the company hired its first AI welfare researcher, Kyle Fish, to help shape early thinking on this issue. Fish is now leading the new research program and has previously stated that there might be around a 15% chance that models like Claude could already be conscious.
In its announcement, Anthropic was clear that there is no scientific consensus on whether current or future AI systems are—or could ever be—conscious. Still, the company believes it's important to approach the topic with humility.
“We’re approaching the topic with as few assumptions as possible,” Anthropic wrote. “We expect our ideas to evolve as the science and technology progress.”
Looking Ahead
Whether or not AI will ever become conscious remains uncertain. But by launching this program, Anthropic is acknowledging a future where these questions might no longer be hypothetical. It’s a thoughtful step into a complex and often controversial area—one that may shape how we view and interact with AI in the years to come.
Related Stories:
Top Stories:
WhatsApp Introduces 'Advanced Chat Privacy' to Strengthen Message Security
WhatsApp is stepping up its commitment to privacy with the introduction of a new feature called Advanced Chat Privacy. This update adds an extra layer of protection to your conversations, helping users feel more secure when discussing sensitive topics.
What Does It Do?
With Advanced Chat Privacy turned on, both you and the people you're chatting with won’t be able to export messages or automatically download media from the chat. It also blocks the ability to mention Meta AI or ask it questions in those conversations.
The goal is to make sure everything said or shared in a chat stays private—especially in situations where trust matters, like in support groups or when you're organizing something meaningful with people you might not know personally.
Why It Matters
While WhatsApp already uses end-to-end encryption to protect your messages, this new feature takes things a step further. It's ideal for conversations that need a higher level of confidentiality, like health-related discussions or community organizing efforts.
How to Turn It On
Enabling Advanced Chat Privacy is straightforward:
- Open the chat
- Tap the name at the top
- Select Advanced Chat Privacy
Once it's turned on, the extra privacy settings apply to that specific conversation.
Still a Work in Progress
Right now, the feature stops users from exporting chats or saving media automatically, but it doesn't block screenshots just yet. WhatsApp has said it plans to expand the feature over time, and screenshot blocking may be included in a future update.
When Will You Get It?
The feature is rolling out gradually over the next few months, so if you don’t see it yet, it should be available to you soon. It will work in both one-on-one chats and group conversations.
Conclusion
Advanced Chat Privacy gives users even more control over what happens in their chats, offering peace of mind when it matters most. Combined with features like disappearing messages and chat lock, it's clear that WhatsApp is continuing to focus on privacy in meaningful ways.
Related Stories:
Top Stories:
