Atlas Awakened

/3 min read
OpenAI unveils a ChatGPT-embedded AI browser to take on Google
Atlas Awakened
(Illustration: Saurabh Singh) 

It has been evident for some time that internet browsers are going to be the new battleground for AI startups and established players like Google and Microsoft. AI startups are looking to break the monopoly the likes of Google enjoy over browsers and search engines by developing their own AI-powered browsers, and the latter are responding by incorporating more generative AI into their platforms. Perplexity created a stir recently when it released a free version of its AI browser Comet. And now OpenAI has just launched its own browser—again free — called ChatGPT Atlas.

What is different about Atlas is that it is built around ChatGPT. Instead of searching for something through a search engine, Atlas takes you to the chatbot. ChatGPT answers or directs you to where you need to go, essentially removing search engines from the process and becoming a new portal to the internet. The browser also has an agent mode (available only to ChatGPT’s Plus and Pro subscribers) that can perform tasks, from booking flight tickets to editing a document that you’re working on.

Atlas is currently available for MacOS computers, although versions for Microsoft Windows, Android and iOS phones are expect­ed later. It’s still early days, but we are probably getting a look into what the browsers of the future might look like.

Virtual Reality

 Samsung has launched Galaxy XR, a mixed-reality headset aimed to compete with Apple’s Vision Pro. Launched in the US and South Korea for now and co-developed with Google and Qualcomm, it deliv­ers immersive virtual environments where apps and content float in the user’s field of view. Users can interact with it through eye tracking, voice commands or hand gestures. It is priced $1,800, about half the cost of Apple’s Vision Pro.

Combating Deepfakes

YouTube has announced that its likeness detection tool, launched earlier as part of a pilot project to protect creators against generative AI content, will become avail­able to more creators. This tool allows people to find and report unauthorised uploads using their likeness. Deepfakes are increasingly becoming a concern, and creators will be able to review videos flagged in the Content Detection tab, and request for removals if videos appear unauthor­ised or AI-generated.

AI Trading Clash

A unique experiment is underway to find which AI model is the soundest at investing. Launched by US research firm Nof1, six top LLMs were given $10,000 each to invest in six cryptocurrency perpetual contracts on Hyperliquid. The models’ objective is to maximise risk-adjusted returns. As of October 22, DeepSeek Chat V3.1 is performing the best with profits of 10.42 per cent. OpenAI’s GPT-5 was the worst with losses of 65.15 per cent.