Weekly AI News

Hey it’s Jul,
Greetings and welcome to the second edition of “Weekly AI News”!

This week has been eventful for OpenAI. First, the company responded to DeepSeek by launching its new reasoning model, o3-mini, on Friday—a significant step forward in the efficiency and speed of language-processing AIs. OpenAI is also rolling out ChatGPT Gov to provide government agencies with secure access to advanced technologies. Strategic collaborations, such as those with National Labs to enhance nuclear security, are taking shape.

Meanwhile, industry leaders like Google and Meta are making notable strides in their own fields. Google is expanding its AI capabilities with experimental features that streamline call handling and enhance user interactions, while Meta is heavily investing in AI infrastructure and personalisation features to push the boundaries of social and digital experiences.

The ongoing debate over the rights to AI-generated content underscores the challenges that lie ahead. As technology rapidly evolves, finding a fair balance between protecting intellectual property and fostering innovation remains a pressing and complex issue for the entire industry.

🚀🤖 OpenAI Launches Free o3-mini Reasoning Model

OpenAI has released its new o3-mini reasoning model for free—a significant leap in speed, accuracy, and cost-efficiency compared to its predecessor. Now integrated into ChatGPT and available via API, the model allows Pro users unlimited access and offers free users a trial through the Reason button. Additionally, a higher-intelligence variant is available for deeper responses.

🏛️🔒 OpenAI Unveils ChatGPT Gov Platform

OpenAI has launched ChatGPT Gov, a specialised version of ChatGPT designed for U.S. government agencies. Hosted on Microsoft Azure and compliant with strict cybersecurity standards, the platform streamlines tasks like automation, coding, translation, and research. It already supports thousands of government employees across multiple agencies.

🔬🛡️ OpenAI Partners with National Labs on Nuclear Security

In a strategic move, OpenAI is partnering with U.S. National Laboratories to grant 15,000 government scientists access to its advanced AI models. This collaboration includes deploying an AI model on Los Alamos’ Venado supercomputer and offering expert guidance on nuclear safety. The initiative aims to integrate AI into high-stakes national security research.

🎨🖥️ ChatGPT Canvas Receives Major Update

ChatGPT Canvas now supports HTML and React code rendering and remains compatible with the o1 model. This long-awaited update, available to Plus members, sets the stage for the upcoming o3-mini release and enhances the platform’s ability to build quick, on-the-fly applications. The improvements mark a significant enhancement in user experience.

💰📈 ChatGPT Pro Subscriptions Drive Revenue Surge

The premium ChatGPT subscription tier, priced at $200 per month, is generating impressive revenue—reportedly at least $25 million monthly. Launched only seven weeks ago, this subscription model is expected to contribute significantly toward OpenAI’s projected $12 billion revenue in 2025. Its rapid growth underscores strong market demand for advanced AI capabilities.

🚨🔍 OpenAI Alleges DeepSeek Used Its Data Illicitly

OpenAI claims that Chinese startup DeepSeek illegally used ChatGPT outputs to train its competing AI model via a distillation process. DeepSeek’s V3 model was built with substantially fewer resources than GPT-4, raising serious concerns over data misuse. Measures have since been taken by OpenAI and Microsoft to prevent further unauthorised use.

📊👥 ChatGPT Subscribers Nearly Triple to 15.5 Million

ChatGPT’s paid subscribers surged from 5.8 million to 15.5 million within a year, contributing to an estimated annualised revenue of around $4 billion. Despite fierce competition from other chatbot providers, this explosive growth solidifies ChatGPT’s market dominance. The surge reflects rapid user adoption and strong demand for its advanced features.

💼💸 OpenAI in Talks with SoftBank at $300B Valuation

OpenAI is reportedly in discussions to raise $40 billion in new funding, potentially elevating its valuation to approximately $300 billion. SoftBank is expected to lead the round with an investment ranging from $15 to $25 billion. This funding effort is driven by the strong momentum behind ChatGPT subscriptions and the overall surge in AI innovation.

☎️🤖 Google’s New AI Manages Phone Calls

Google has introduced two experimental features in Search Labs that handle phone calls on behalf of users. “Ask for Me” contacts local businesses to gather service details, while “Talk to a Live Representative” manages customer service calls by alerting users when a representative is available. Powered by Duplex technology, these features aim for natural and efficient voice interactions.

🔍🚀 Google Unveils Next-Gen Flagship AI Model

Google quietly announced its new flagship AI model, Gemini 2.0 Pro Experimental, via its Gemini chatbot app. Engineered to enhance factual accuracy and improve coding and mathematical performance, the model is available in early preview for advanced users. This release represents a pivotal step in Google’s ongoing AI evolution.

💼🌐 Zuckerberg Announces $65B AI Investment Plan

Meta CEO Mark Zuckerberg has unveiled a bold $60–65 billion investment plan in AI infrastructure for 2025. The plan includes deploying 1GW of compute power and acquiring over 1.3 million GPUs, marking one of the largest AI hardware rollouts globally. This strategic investment is designed to position Meta AI as a leader in digital assistant technology.

👤✨ Meta AI Gets a Personalisation Upgrade

Meta has enhanced its AI assistant with new personalisation features that allow it to remember user details and access data across Facebook, Instagram, and WhatsApp. The upgraded assistant delivers tailored recommendations and responses based on individual user profiles and past interactions. However, concerns about privacy have emerged due to the lack of opt-out options.

🖼️🔥 DeepSeek Launches New AI Image Model Janus-Pro

Chinese startup DeepSeek has unveiled Janus-Pro, an open-source multimodal AI image generation model that outperforms competitors like DALL-E 3 and Stable Diffusion. Available in both 1B and 7B parameter configurations, Janus-Pro excels in quality and accuracy benchmarks. Released under an MIT license, it offers developers extensive freedom for commercial projects.

🔓📂 DeepSeek Database Exposes Over 1M Log Lines and Secret Keys

A misconfigured ClickHouse database at DeepSeek left over one million lines of logs, secret keys, and sensitive backend details exposed online. Security researcher Gal Nagli warned that the breach could allow full control over internal operations. DeepSeek promptly secured the database following alerts from cloud security experts.

🇮🇹🚫 Italian Regulator Blocks DeepSeek AI App Over Data Concerns

Italy’s data protection authority, the Garante, has ordered DeepSeek to block its chatbot in the country due to concerns over its privacy policy and data handling practices. The regulator scrutinised how personal data is collected, stored, and used, particularly with respect to sensitive information. This decision reflects growing European oversight of Chinese tech firms.

🕵️‍♂️🔓 Perplexity Offers Uncensored Access to DeepSeek AI Search

Startup Perplexity now provides uncensored access to DeepSeek R1 by removing content moderation filters. This change enables users to obtain unfiltered answers, even on sensitive topics such as the 1989 Tiananmen Square protests. The move emphasises Perplexity’s commitment to open inquiry and transparency in AI search.

🤝📱 Perplexity AI Proposes New U.S. Merger with TikTok

Perplexity AI has revised its merger proposal for TikTok’s U.S. operations, suggesting the formation of a new entity called “NewCo” that could be valued up to $300 billion post-IPO. The proposal also includes a plan for the U.S. government to acquire up to a 50% stake. This ambitious strategy has captured the interest of major tech players and could reshape the social media landscape.

⚙️📏 Qwen Launches Model Upgrades with 1M Token Support

Alibaba’s Qwen team has released the Qwen2.5-1M series, featuring models with 7B and 14B parameters that support context lengths of up to 1 million tokens. Utilising a custom vLLM inference framework, these models operate up to seven times faster than traditional long-context systems. The update also introduces Qwen Chat v0.2 with enhanced capabilities like web search and text-to-video generation.

📲🖼️ Qwen Releases AI Models Capable of Device Control

Alibaba’s Qwen team has unveiled Qwen2.5-VL, a family of vision-language models that can interact with computers and smartphones. The flagship 72B model excels at complex tasks such as document parsing and video analysis and can even perform actions like booking flights and editing images. Lighter versions are available freely, while the 72B model requires permission for extensive commercial use.

💻📉 Microsoft Reports Slowing Revenue Growth, but AI Sales Soar

Microsoft announced a 12% revenue growth in Q4 2024, slightly below previous quarters due to constraints in scaling Azure’s server capacity. However, AI-related product sales are booming, with annualised AI revenue now projected to exceed $13 billion. This highlights the strong momentum in Microsoft’s AI segment despite broader challenges.

🆓🔍 Microsoft Makes Copilot’s Think Deeper Feature Free for Everyone

Microsoft has made its Think Deeper mode, which provides detailed step-by-step responses to complex queries, available to all Copilot users. Previously limited to Copilot Pro subscribers, this feature is powered by OpenAI’s o1 reasoning model. The decision removes a $20 monthly barrier, democratising advanced AI assistance for a wider audience.

🚀👀 xAI’s Grok-3 Model Briefly Goes Live

xAI’s upcoming Grok-3 model made a brief appearance for select users, showcasing enhanced reasoning capabilities. Although still in the experimental phase, the demo has generated considerable interest among early adopters. A full public release is expected later this week, promising further improvements in performance and reliability.

🎥✨ Pika Labs Releases v2.1 of Its AI Video Generation Model

Pika Labs has launched version 2.1 of its AI video generation model, featuring advanced motion control, realistic physics, and customisable scene elements. The update delivers 1080p resolution with smoother movements, enhanced character coherence, and improved text alignment. This upgrade paves the way for creating more realistic and dynamic video content using AI.

⏳🔮 Anthropic CEO Predicts AI Could Double Human Lifespan by 2030

Anthropic CEO Dario Amodei has predicted that AI advancements could potentially double human lifespans by 2030. He envisions that AI may condense a century’s worth of research breakthroughs into just 5 to 10 years. This bold projection has ignited debate over the transformative potential of AI on health and longevity.

📝🤔 Anthropic CEO Comments on DeepSeek’s Progress and U.S. Chip Controls

In a recent essay, Anthropic CEO Dario Amodei weighed in on DeepSeek’s R1 release amid growing U.S. chip export controls. He argued that DeepSeek’s advancements align with expected industry cost reductions rather than constituting a revolutionary breakthrough. His analysis raises important questions about the sustainability of rapid AI progress under current hardware restrictions.

🎶🔊 Riffusion Launches Free AI Music Platform “Fuzz”

San Francisco startup Riffusion has unveiled Fuzz, a free AI music platform that creates full-length songs from text prompts, audio snippets, or images. The adaptive platform learns from users’ musical preferences to generate personalised tracks over time. Backed by advisory support from The Chainsmokers, Fuzz is set to disrupt the music creation landscape.

💡🎗️ Reid Hoffman Launches Manas AI to Tackle Cancer

LinkedIn co-founder Reid Hoffman is stepping into healthcare with his new startup Manas AI, which recently raised $25 million from General Catalyst. The company aims to leverage AI to accelerate the discovery of innovative cancer treatments. This initiative underscores the expanding role of AI in addressing complex medical challenges.

🚗🤖 Elon Musk Aims to Rebrand Tesla as an AI Company

Elon Musk is shifting Tesla’s focus from traditional car manufacturing to positioning itself as an AI-driven company. In a recent earnings call, Musk emphasised Tesla’s ventures into autonomy, robotics, and AI-powered technologies over its core automotive business. This rebranding has left investors questioning the company’s long-term strategic direction.

🐥💻 Block Releases Open-Source AI Agent Platform “Goose”

Jack Dorsey’s fintech company, Block, has launched Goose, an open-source framework for building and deploying AI assistants across multiple platforms. The platform supports various LLM backends, including OpenAI, DeepSeek, and Anthropic, while ensuring data privacy and flexible integration. Goose aims to simplify complex digital tasks, similar to Block’s innovations in digital payments.

🎼🔓 YuE Unlocks Open-Source AI Music Generation

Researchers at the Hong Kong University of Science and Technology have introduced YuE, an open-source AI music generation tool that transforms lyrics into full-length songs. By leveraging two specialised models—one for vocals and another for production—YuE can produce tracks up to five minutes long in multiple languages. This breakthrough challenges proprietary music platforms and opens new creative avenues.

📚⚖️ U.S. Copyright Office Sets Clear Rules on AI-Generated Works

The U.S. Copyright Office has released a 52-page report clarifying that works generated solely by AI are not eligible for copyright protection. Only works that combine significant human creativity with AI output qualify for legal protection. This guideline offers much-needed clarity for creators and companies operating in the evolving landscape of AI art.

🤖💃 Unitree Robots Master Traditional Chinese Dance

At China’s Spring Festival Gala, Unitree demonstrated 16 humanoid robots performing traditional folk dances alongside human performers. Using advanced AI motion control and 3D laser SLAM technology, the robots executed complex dance routines with impressive precision. This display highlights significant progress in coordinated human-robot interaction and movement.

⏰🌍 Doomsday Clock Moves to a Record 89 Seconds to Midnight

The symbolic Doomsday Clock has been set to a record 89 seconds before midnight, reflecting escalating global threats. Experts point to factors such as AI-powered military operations, climate change, nuclear proliferation, and geopolitical instability. This stark indicator underscores the urgent need to address the multifaceted risks facing our world today.

Community

Join AI Whisperer Community!

Ready to take your AI journey to the next level? Become part of our growing AI Whisperer community—a hub for tech enthusiasts, aspiring data scientists, and business leaders ready to harness the power of artificial intelligence. Inside, you’ll find:

• Community: Like-minded individuals keen to grow together

• Curated AI Tools: Discover the latest and greatest tools to boost productivity.

• Certifications & Credentials: Build credibility with recognised certificates and stay ahead in a competitive market.

Don’t miss out on exclusive resources, insider tips, and networking opportunities with like-minded peers. Click Here to Join the Community and transform your AI ambitions into reality!

That's it for this week!
Until next time, stay curious and keep exploring the ever-evolving world of AI!
Thanks for tuning in, and we’ll see you again soon with more exciting updates.

Jul