Long story short: OpenAI isn’t sitting on its hands. Shortly after the DeepSeek ruckus shook the industry, they announced the new o3-mini model—a clear statement of power that outperformed DeepSeek’s R1. Then, OpenAI surprised everyone with Deep Research, an AI agent that autonomously gathers, reasons, and synthesizes information from countless sources. And as if that wasn’t enough, SoftBank has now stepped up with talks of a $15–25 billion investment and the launch of SB OpenAI Japan.
Imagine having an AI that goes off and does all the hard work of researching a topic for you—gathering data, cross-referencing multiple sources, and pulling together clear, synthesized answers. OpenAI’s new Deep Research initiative is set to do just that, revolutionizing how we access knowledge.
What Is Deep Research? Deep Research is OpenAI’s latest project designed to accelerate scientific discovery and innovation. But it’s not just for scientists—this tool is built to help anyone who needs reliable, in-depth research without the hassle of spending hours scouring the web.
If you’re looking for a smarter way to search the internet, Perplexity AI is leading the way. Recently, Perplexity Pro has been upgraded with two advanced reasoning models that make searching, researching, and planning information a breeze.
DeepSeek R1: Open-Source Power and Uncensored Processing Integration Date: January 28, 2025
Perplexity Pro kicked off its upgrade by integrating DeepSeek R1—a fully open-source reasoning model hosted on US servers. This move addresses some of the common security and censorship concerns you might face with using DeepSeek directly on its own Chinese hosted website.
Imagine an internet search engine that doesn’t just return a list of links and ads but instead converses with you—finding exactly what you need by pulling together information from across the web. That’s Perplexity, and now you can get the paid for Perplexity Pro edition for free through an exclusive offer.
What Is Perplexity Pro? Perplexity is a next-generation AI-powered search engine that goes beyond traditional keyword searches. Instead of you having to sift through countless ads and irrelevant results, Perplexity gathers and distills information from across the internet for you into clear, conversational answers.
Microsoft’s latest documentation on Microsoft 365 Copilot reveals some critical insights for anyone using SharePoint as a knowledge source for their AI agents. These details clarify how Copilot processes content and highlight important best practices to ensure your data plays nicely with the system.
Key Insights from the Documentation File Count Matters:
5 or Fewer Files:
When you add up to 5 SharePoint files as knowledge sources, Copilot uses the full content of these files in its responses.
Sam Altman’s recent Reddit AMA (ask me anything which happened on 31st January 2025) has stirred conversation across the AI community. In response to news that competitors DeepSeek and Qwen are giving open-source access to their large language models, Altman made it clear: competition is exactly what the industry needs.
A Positive Spin on Open Access Altman’s core message was straightforward. He welcomed DeepSeek and Qwen’s decision to let users download and experiment with their LLMs.
In the rapidly evolving field of artificial intelligence, China’s latest entrant, Qwen AI, is making notable progress. Developed by Alibaba Cloud, Qwen AI demonstrates that generative AI models can be built both affordably and efficiently while offering a range of creative and practical applications.
Advanced Features Qwen AI is more than just a text-based chatbot. It also supports:
Image Generation: Users can generate images by simply providing descriptive prompts within the chat interface.
Here’s something exciting for the weekend: OpenAI has just released its latest reasoning models, o3-mini and o3-mini-high. This is the first time advanced reasoning capabilities are available to free ChatGPT users — an important milestone in making sophisticated AI more accessible. On top of this, unlike o1 that came before it, the new o3-mini reasoning model works alongside “search” to be able to access live results from the internet.
Introduction to o3-mini The o3-mini model is a streamlined version of OpenAI’s o3 reasoning series.
AI startup DeepSeek recently suffered a serious security breach involving a misconfigured ClickHouse database. The database—left accessible via open ports 8123 and 9000 on two subdomains—lacked authentication, exposing over a million lines of sensitive information to the public internet. This incident raises important questions about how emerging AI companies balance rapid growth with the need for robust security measures.
Key Details of the Breach Publicly Accessible Database DeepSeek’s ClickHouse instance allowed anyone with the correct URL to execute arbitrary SQL queries through its HTTP interface.
DeepSeek’s rapid rise in the AI industry has brought both applause for its cost-effective approaches and skepticism regarding its transparency. While its AI assistant soared to the top of app store charts in both the UK and US, questions remain about how accurately DeepSeek represents its training costs, funding sources, and technological achievements. Meanwhile, new announcements—like DeepSeek’s omnimodal Janus-Pro-7B model—demonstrate the company’s ambitions aren’t slowing down.
Cost and Funding Controversies Training Costs Under $6 Million?
DeepSeek, a Chinese startup, has made headlines by releasing an AI assistant that quickly rose to the top of both the UK and US Apple App Store charts. This surge in popularity has placed it above established names like ChatGPT, sparking conversations around innovation, cost-effective AI development, and privacy concerns.
Key Highlights from DeepSeek’s Approach Budget-Friendly Development DeepSeek’s DeepSeek-V3 model was developed for under $6 million, which is far less than many competitors spend.
OpenAI recently stirred the AI community with news that its o3-mini model will soon be accessible to a broader audience. While official details remain somewhat sparse, we have enough context—and a few recent updates—to get a clearer picture of what o3-mini might offer, how it builds on o1, and what to expect from this new addition to the OpenAI ecosystem.
Quick Highlights: New Info as of January 17, 2025 CEO Announcement: Sam Altman confirmed o3-mini is finalised and set for release in the coming weeks.
The Canvas feature in ChatGPT has been turning heads since it first launched, especially among developers who want a quick, interactive way to test and display code. Until recently, though, Canvas was only available on GPT-4o—which not everyone had access to. Now, it’s rolled out to o1, ChatGPT’s more advanced reasoning model, opening the door for a wider group of users to benefit from real-time HTML and React rendering. Let’s break down why this matters and what you can do with it.
If you’ve ever used an AI tool and wondered, “Where’s this information coming from?”, you’re not alone. Tools like ChatGPT, Perplexity, and NotebookLM already offer some form of source referencing, but Anthropic’s new Citations feature for Claude takes things a step further in terms of precision, flexibility, and deep integration. Let’s explore what makes this development so significant—and how it sets a new standard for trustworthy AI.
Overview: Grounding AI Responses in Real Sources Anthropic has rolled out a Citations feature for its Claude AI models—currently Claude 3.
Another week, another leap forward in the AI space! This time it’s Perplexity making waves with their brand-new AI Assistant for Android. If you’re wondering what it does, why it matters, and how it compares to what’s already out there, let’s break it down in a straightforward way.
What Is Perplexity’s Android Assistant? Perplexity has taken its AI capabilities and wrapped them into a handy new assistant designed specifically for Android devices.
There’s been a lot of buzz around OpenAI’s latest feature, Operator. If you’re wondering what it is, what it can do, and why people are so excited, don’t worry—you’re not alone. At first glance, it might sound a bit like something out of a sci-fi film, but really, it’s a clever step forward in how we use AI to simplify tasks. Let’s break it down.
What Is Operator? Simply put, Operator is like having an AI personal assistant that can use the internet on your behalf.
If there’s one thing you can count on in the fast-paced world of tech, it’s that nothing stays the same for long. This past week has been a whirlwind for us—wrapping up our first Generative AI Skills Bootcamp cohort, hosting back-to-back events, and then watching in disbelief as five major AI announcements dropped just hours after we’d updated everyone on the latest news. Let’s dive in!
Monday: Wrapping Up Our First Generative AI Skills Bootcamp Ellie and I run the Generative AI Skills Bootcamp together, and on Monday, we watched our first cohort deliver their final business plan presentations.
The tech to build the holodeck
A visit to an elderly mother in Germany highlighted the capability of Gaussian splatting, a novel 3D capture technique creating lifelike 3D models with your phone. This technology, developed in 2023 by European scientists, uses millions of fuzzy, translucent blobs to capture the precise color, location, scale, and transparency of objects, enabling photorealistic 3D reconstructions. Major tech firms like Niantic, Google, Snap, and Meta are utilizing this breakthrough for various applications, from mapping and geospatial games to volumetric video capture and metaverse spaces.
This weekend has been nothing short of extraordinary for TikTok users in the United States. Just this morning, on January 19th, TikTok went dark, leaving millions of users scrambling to adapt and countless small businesses in disarray. Yet, in an astonishing turn of events, the app is now being restored following an intervention by President-elect Donald Trump. All of this has happened within 24 hours — and during a weekend, no less.
As of January 19, 2025, TikTok has officially gone dark in the United States. The popular app, known for its viral dance trends, comedy sketches, and unparalleled ability to catapult creators into stardom, has ceased operations in anticipation of an impending federal ban. For millions of Americans, this marks the end of an era — and the start of a scramble to adapt.
Why Did TikTok Shut Down in the U.