Get the latest AI news, understand why it matters, and learn how to apply it in your work — all in just 5 minutes a day. Join over 2,000,000+ subscribers.
OpenAI reveals ChatGPT agent for ‘deep research’
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. The AI transition from instant answers to complex reasoning just leveled up again, with OpenAI’s new feature putting even the hardest benchmarks on notice.
With ‘Deep Research’ autonomously crushing complex research tasks in minutes, the evolution of how we use AI assistants just took another major leap forward.
In today’s AI rundown:
OpenAI debuts AI research assistant
OpenAI introduces o3-mini reasoning model
How to build your own free AI Operator
Sam Altman’s stance on open source
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
OPENAI
🔍 OpenAI debuts AI research assistant
Image source: OpenAI
The Rundown: OpenAI just launched Deep Research, a new ChatGPT feature that conducts extensive web research on complex topics and delivers detailed reports with citations in under 30 minutes.
The details:
The system uses a specialized version of o3 to analyze text, images, and PDFs across multiple sources, producing comprehensive research summaries.
Initial access is limited to Pro subscribers ($200/mo) with 100 queries/month, but if safety metrics remain stable, it will expand to Plus and Team users within weeks.
Research tasks take between 5-30 minutes to complete, with users receiving a list of clarifying questions to start and notifications when results are ready.
Deep Research achieved a 26.6% on Humanity’s Last Exam, significantly outperforming other AI models like Gemini Thinking (6.2%) and GPT-4o (3.3%).
Why it matters: ChatGPT excels at quick, instant answers, but Deep Research represents the first major consumer attempt at tackling complex tasks that take humans days. Combined with the release of Operator, the landscape is shifting towards longer thinking with autonomous actions — and better results to show for it.
TOGETHER WITH VANTA
🛡️ Live event: AI & security maturity
The Rundown: Join John Hammond — cybersecurity researcher, practitioner, and content creator — and Matt Cooper, Vanta’s Director of GRC, for a fireside chat on AI, security maturity, and the top security risks in 2025.
What you’ll gain:
A deep dive into 2025’s top cyber risks, including the impact of AI
Actionable insights to refine your security priorities
Strategies tailored to your organization’s security maturity level
A live Q&A at the end
Save your spot and get ready to future-proof your approach to cybersecurity.
OPENAI
🧠 OpenAI introduces o3-mini reasoning model
Image source: OpenAI
The Rundown: On Friday, OpenAI released o3-mini, a new cost-efficient reasoning model that brings advanced STEM capabilities to both free and paid users while significantly reducing costs and response times compared to previous versions.
The details:
Free users can access reasoning capabilities for the first time via o3-mini, with paid users getting expanded rate limits of up to 150 messages daily.
o3-mini shows particular strength in technical domains like math and coding, matching or exceeding o1's performance while responding 24% faster.
Developers can also fine-tune the ‘reasoning effort’ with three settings — low, medium, and high — to balance speed and accuracy for different use cases.
The system costs 63% less to run than its predecessor, bringing the price down to $1.10 / million input tokens while maintaining competitive performance.
Why it matters: DeepSeek has dominated the headlines over the past week, but OpenAI always seems to have another ace up its sleeve. o3-mini will expose the free tier of AI users to their first ‘reasoning’ experience—and with the full o3 coming in ‘less than a few months,’ the next step up in the model cycle is just around the corner.
AI TRAINING
🤖 How to build your own free AI Operator
The Rundown: With this tutorial, you can run a powerful AI browser automation tool on your computer that can control web browsers and perform tasks autonomously without paying for expensive subscriptions like ChatGPT’s $200/month plan.
Step-by-step:
Download Ollama and install a language model (e.g., deepseek-r1:14b or qwen2.5:7b).
Clone the web interface repository with git clone. (https://github.com/browser-use/web-ui.git)
Set up Python environment and install dependencies.
Launch the interface, configure your AI assistant, and see it perform web actions for you.
Pro tip: Start with basic tasks to understand the system's capabilities before moving to complex automation. We just did a full workshop on how to set it up properly, which you can access here.
PRESENTED BY INNOVATING WITH AI
💼 Start your career as an AI Consultant
The Rundown: Innovating with AI’s new program, AI Consultancy Project, equips AI enthusiasts with all the resources to capitalize on the rapidly growing AI consulting market – which is set to 8x to $54.7B by 2032.
The program offers:
Tools and framework to find clients and deliver top-notch services
A 6-month roadmap to build a 6-figure AI consulting business
Student landing their first AI client in as little as 3 days
Click here to request early access to The AI Consultancy Project.
OPENAI
🗣️ Sam Altman’s stance on open source
Image source: Reddit
The Rundown: OpenAI held a Reddit AMA following the launch of its o3-mini model, with CEO Sam Altman, CPO Kevin Weil, and others tackling questions ranging from candid commentary on open-source tech to upcoming feature and model timelines.
The details:
Altman revealed that he personally thinks ‘we have been on the wrong side of history’ regarding open source and that OpenAI needs to figure out a strategy.
The CEO also said that he guessed the full o3 would be released in ‘more than a few weeks, less than a few months.’
Altman also commented on DeepSeek, calling it a ‘very good model’ and saying OpenAI will ‘produce better models, but maintain less of a lead.’
Weil provided a few nuggets about upcoming releases, saying more agents and a new image generator will be available very soon.
Why it matters: The biggest headline is the open-source commentary, which is the biggest criticism OpenAI faces. While no changes seem imminent (Altman said not everyone at the company shares his opinion, and it's not a high priority), a shift in any capacity would be a paradigm-altering moment for the entire industry.
QUICK HITS
🛠️ Trending AI Tools
☎️ Ask For Me - Google’s AI that calls businesses on your behalf
📞 Talk to a Live Representative - Google’s AI that waits on hold for you
🤖 Mistral Small 3 - New fast, powerful 24B parameter open-source model
🎶 Riffusion - Create full songs for free from text, image, or audio prompts
💼 AI Job Opportunities
🎯 The Rundown - Marketing/Media Buyer
⚙️ Palantir Technologies - Product Manager
💡 Curai - Solutions Consultant
🏗️ OpenAI - Data Center Design Engineer
📰 Everything else in AI today
U.S. AI czar David Sacks shared a new report estimating DeepSeek has spent over $1B on computing, calling the $6M training cost number ‘highly misleading.’
The EU activated the first phase of its AI Act, banning AI systems deemed ‘unacceptably risky’ and imposing penalties of up to $35M euros.
Google’s X moonshot lab launched Heritable Agriculture, an agriculture company using AI and machine learning to accelerate plant breeding for improved crop yields.
Microsoft AI CEO Mustafa Suleyman announced a new cross-disciplinary research unit, recruiting economists, psychologists, and others to study AI's societal impact.
MIT researchers unveiled ChromoGen, an AI model that predicts 3D genome structures in minutes instead of days and enables DNA analysis and how if impacts cell function and disease.
Security researchers discovered an exposed DeepSeek database containing over 1M user prompts and API key records, raising vulnerability and privacy concerns.
COMMUNITY
🎥 Join our next live workshop
Join our next advanced workshop on Tuesday at 3 PM EST for a special 90-minute deep dive into building revenue-generating mobile apps with Kevin Hou, Head of Product Engineering at Codeium.
RSVP here. Not a member yet? Join The Rundown University on a 14-day free trial.
🤝 Share The Rundown, get rewards
We’ll always keep this newsletter 100% free. To support our work, consider sharing The Rundown with your friends, and we’ll send you more free goodies.
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
OpenAI goes nuclear
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. OpenAI has been making plenty of moves with the U.S. government, but it might’ve just been handed its most sensitive assignment yet — helping protect the nation's nuclear arsenal.
With the company’s models now deployed across National Labs for nuclear science, physics, and disease research, AI is quickly becoming a national security cornerstone.
Reminder: Our next workshop is today at 4 PM EST! Learn about how to automate tasks with OpenAI’s Operator and create your own free, personalized, offline assistant.
In today’s AI rundown:
OpenAI teams up with National Labs on nuclear security
Google’s new AI handles phone calls
Create AI visuals containing legible text
Riffusion launches free AI music platform
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
OPENAI
☢️ OpenAI teams up with National Labs on nuclear security
Image source: Reuters
The Rundown: OpenAI just announced a new partnership with U.S. National Laboratories, giving thousands of government scientists access to its most advanced AI models for critical research, including nuclear weapons security.
The details:
15k scientists will gain access to o1 models, with OpenAI supporting research including cybersecurity, power grid protection, disease treatment, and physics.
The company will also deploy an AI model on Los Alamos' Venado supercomputer in partnership with Microsoft.
OpenAI researchers with security clearances will consult on nuclear security projects focused on weapons safety and nuclear war risk reduction.
The partnership follows OpenAI's recent release of ChatGPT Gov, a specialized platform for government use across federal agencies.
Why it matters: AI systems are being integrated into the U.S.’s most sensitive national security infrastructure, and OpenAI continues to position itself as a crucial part of the country’s tech development. But if AI’s scientific capabilities reach even a fraction of the level many expect, it will become a global government necessity, not just a priority.
TOGETHER WITH HUBSPOT
⛏️ The AI gold rush: Are you in or out?
The Rundown: HubSpot’s ‘AI For Business Builders’ free guide shows you how to join the savvy entrepreneurs quietly leveraging artificial intelligence for real business growth — with no P.H.D. needed.
This no-BS guide cuts through the hype, giving you:
A jargon-free breakdown of large language models (the engine behind AI's magic)
Concrete steps to leverage AI for immediate business growth
Real-world examples that turn "prompt engineering" from gibberish into a secret weapon
📞 Google’s new AI handles phone calls
Image source: Google Labs
The Rundown: Google just released two new Search Labs experimental features that handle phone calls on a user's behalf, including ‘Ask for Me’ for inquiring about local services and ‘Talk to a Live Representative’ for customer service calls.
The details:
‘Ask for Me’ can contact local businesses to gather pricing and availability information for services like auto repairs and nail salons.
Users enter requirements through a search interface, with Google's AI handling the phone call and providing a summary via text or email within 30 minutes.
A separate ‘Talk to a Live Representative’ feature waits on hold with customer service lines and alerts users when a representative is available.
Both features utilize Google's advanced Duplex AI technology for natural-sounding voice interactions.
Why it matters: With much of the younger generation hating phone interactions, these new Google features could have major mass appeal. With much of customer service also transitioning to artificial intelligence, the future is (for better or worse) likely looking like a wave of AIs calling other AIs.
AI TRAINING
🖼️ Create AI visuals containing legible text
The Rundown: Ideogram turns text into eye-catching images for free, with a special ability to incorporate legible text into the generated visuals.
Step-by-step:
Sign up for a free account at Ideogram (you'll get free daily generations).
Click "What do you want to create?" on the dashboard and craft a detailed prompt.
Adjust settings like aspect ratio and magic prompts and hit "Generate" to get four unique images.
Pro tip: After generating your images, use "Remix" for subtle tweaks, "Retry" for new variations, or "Upscale" for higher resolution.
PRESENTED BY SECTION
🔑 Unlocking personalized AI at scale
The Rundown: Join Jan Pedersen, Walmart’s VP of Search & Personalization, and Greg Shove, Section CEO for an in-depth look into how companies can implement AI-driven personalization at scale.
Join the free session on Feb. 5 to learn:
How AI is transforming the way businesses deliver personalized experiences
Insights from real personalization strategies at Walmart
Risks and opportunities in implementing AI-driven personalization across industries
RIFFUSION
🎵 Riffusion launches free AI music platform
Image source: Riffusion
The Rundown: San Francisco startup Riffusion just launched Fuzz, a new free generative AI music platform that allows users to create full-length songs and can adapt to a user’s preferences and tastes over time.
The details:
The platform allows users to create full-length original music through simple text prompts, audio snippets, or image inputs.
Fuzz also features an adaptive learning element, allowing the model to learn a user’s musical preferences through their generations and profiles.
The company raised $4M in 2023, with electronic music group The Chainsmokers acting as advisors and testers for the platform.
Why it matters: This has been a big week for free AI music, with the open-source YuE and now Fuzz hitting the scene. With increasing capabilities and lower costs, it's clear there is an AI music revolution coming — and it's likely that many songs in the wild are already leveraging the tech in some capacity without listeners even realizing it.
QUICK HITS
🛠️ Trending AI Tools
🎥 Upscale to 4K - Luma Labs’ new feature to dramatically increase the resolution of a generated video
🧠 Think Deeper - New reasoning capabilities available to all Copilot users
📝 Stella - Automate tasks like meeting invites, emails, and note-taking with AI
📣 Kiva - AI-powered SEO agent for agencies, SMEs & startups
💼 AI Job Opportunities
🤖 The Rundown - Writer (Robotics/Tech)
🧠 OpenAI - Research Engineer/Scientist, Model Fusion
🏠 Waymo - Real Estate Development Manager
💼 Kumo - Account Executive (New York)
📰 Everything else in AI today
OpenAI is reportedly in talks to raise up to $40B at a staggering $340B valuation, potentially more than doubling its worth from late 2024.
Google is rolling out Gemini 2.0 Flash across its mobile and web apps, featuring faster responses, improved image generation through Imagen 3, and enhanced performance.
Krea AI teased Krea Chat, an upcoming tool powered by DeepSeek that provides a text interface for generating and editing images and videos on the platform.
Mistral released Small 3, a 24B-parameter open-source model that matches the performance of 70B models at 3x the speed and is deployable on consumer hardware.
Sakana AI unveiled TinySwallow-1.5B, a new tiny Japanese language model that can run offline on smartphones and achieves top performance among similar-sized models.
ElevenLabs officially announced a $180M Series C funding round, bringing the AI speech startup’s valuation to over $3B.
AI2 introduced Tülu 3 405B, the Seattle-based non-profit’s largest open-source model yet that surpasses DeepSeek V3 and GPT-4o on certain benchmarks.
COMMUNITY
🎥 Join our next live workshop
Join our next workshop this Friday at 4 PM EST to learn about how to automate tasks with OpenAI’s Operator and create your own free personalized, offline assistant with Dr. Alvaro Cintas, The Rundown’s AI professor.
RSVP here. Not a member? Join The Rundown University on a 14-day free trial.
🤝 Share The Rundown, get rewards
We’ll always keep this newsletter 100% free. To support our work, consider sharing The Rundown with your friends, and we’ll send you more free goodies.
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
AI gets new copyright clarity
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. Put down the pitchforks and pick up your AI tools — the U.S. Copyright Office just gave creators the best of both worlds.
With new guidelines that protect human artistry while embracing technological innovation, are we about to witness a broader explosion of AI-enhanced creativity?
In today’s AI rundown:
U.S. Copyright Office sets clear rules on AI
Anthropic CEO weighs in on DeepSeek
Create design-ready brand mood boards
Unitree robots master traditional Chinese dance
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
AI & COPYRIGHT
📋 U.S. Copyright Office sets clear rules on AI
Image source: The U.S. Copyright Office
The Rundown: The U.S. Copyright Office just released a new report establishing firm guidelines on AI-generated works, ruling that AI outputs alone cannot receive copyright protection while preserving rights for human creators who use AI as a tool.
The details:
The 52-page report determined that copyright protection requires meaningful human authorship and creativity, not just AI generation.
Even with extensive prompt engineering, simply providing text prompts to AI systems generally doesn't qualify for copyright protection.
The report highlighted works that combine human-authored elements with AI-generated content as copyrightable, but only for the human-created portions.
The Office also said no new legislation is needed at this time to handle AI copyright issues, with current registration policies continuing as normal.
Why it matters: This guidance provides much-needed clarity for creators and companies working with AI tools while still protecting human authorship. As more artists and businesses continue to grapple with integrating AI into workflows while seeking to protect valuable IP, the ruling comes at a vital time.
TOGETHER WITH HUMANX
🏟️ Double the AI insights at HumanX
The Rundown: HumanX is the premier AI event of 2025, bringing together top-tier speakers and innovators across the industry — and for 72 hours (or until sold out), you can buy one all-access ticket and get one free!
At HumanX, you’ll experience:
300+ speakers from the world’s most influential AI-driven companies
Handpicked networking opportunities tailored to your function or industry
Comprehensive recordings, transcripts, and AI-powered session insights post-event
Lounges, gourmet catering, and sessions that keep you energized, not exhausted
Secure Your Spot Now — take advantage of the 2-for-1 flash deal for the next 72 hours and lock in discounted rooms at the Fontainebleau before they’re gone.
ANTHROPIC
📝 Anthropic CEO weighs in on DeepSeek
Image source: Getty Images
The Rundown: Anthropic CEO Dario Amodei just published a new essay detailing his thoughts on DeepSeek’s R1 release and U.S. AI chip controls, also addressing misconceptions about the Chinese company's achievements.
The details:
He argues that DeepSeek's progress follows expected industry cost reductions and matches the U.S. from months ago rather than a breakthrough.
Amodei revealed that Claude 3.5 Sonnet's training costs were in the "tens of millions," challenging DeepSeek's claimed $6M efficiency advantage narrative.
Looking ahead to 2026-2027, Amodei projects that building superintelligent AI will require millions of chips and tens of billions in investment.
He also said current export controls are impacting DeepSeek's reliance on mixed chip types, suggesting the hardware restrictions are working.
Why it matters: Amodei’s views starkly contrast the hype we’ve seen across the media and global discourse regarding DeepSeek. The competition between the U.S. and China is entering a critical phase, and the industry debate seems to be shifting from whether to restrict chip exports at all to how extensively they should be controlled.
AI TRAINING
🎨 Create design-ready brand mood boards
The Rundown: ChatGPT lets you transform your brand ideas into professionally curated mood boards by getting color palettes, typography pairings, and visual style recommendations.
Step-by-step:
Provide ChatGPT with your brand details and 2-3 reference images of your desired aesthetic.
Request a color palette with hex codes specifying your industry and brand values.
Upload typography examples you like and get Google Fonts pairings.
Share mood board references and ask for detailed guidelines.
Pro tip: Keep your reference images consistent with your brand styles. If you’re a Rundown University member, copy/paste our prompts from here.
PRESENTED BY POSTMAN
🤖 Streamline AI agent development
The Rundown: Postman’s AI Agent Builder brings everything you need — LLMs, APIs, and helpful workflows — into one intuitive platform, making it easier than ever to create reliable, functional AI agents.
With Postman, you can:
Find and use trusted APIs and language models in one central place
Turn APIs into instant agent-ready code
Create workflows using a simple visual editor
Compare performance of leading LLMs side-by-side to choose the right model for your business.
UNITREE
💃 Unitree robots master traditional Chinese dance
Image source: CCTV
The Rundown: Chinese robotics company Unitree just showcased 16 humanoid robots performing traditional folk dances alongside human performers at China's Spring Festival Gala, demonstrating advances in AI-driven motion control.
The details:
The robots utilized AI motion control and 3D laser SLAM technology to execute complex dance moves, such as handkerchief spinning and synchronized leg kicks.
Unitree released a new open-source full-body dataset earlier this month, enabling humanoid robots to achieve more natural, human-like movements.
AI algorithms also allow the robots to "understand" music and adjust their movements to match rhythm and beat in real time.
The Unitree H1 robots also have 360o panoramic depth awareness, allowing for coordinated dance moves and navigation.
Why it matters: The demos keep getting wilder, with Unitree showing major coordination and dexterity in this latest dance routine. The combination of motion control, environmental awareness, and synchronization between 16 humanoids is also a sign we’re entering a new era of capabilities for robots working alongside humans.
QUICK HITS
🛠️ Trending AI Tools
🦢 Goose - Open-source AI agent platform for automating engineering tasks
🤖 Qwen2.5-Max - Alibaba’s new MoE model that outperforms 4o, DeepSeek V3, and Claude 3.5 Sonnet across key benchmarks
🎶 YuE - Open-source lyric-to-music model
🔎 Jasper - Create SEO-optimized content in minutes with AI
💼 AI Job Opportunities
📣 The Rundown - Marketing/Media Buyer
🤝 Findem - Customer Success Manager (Pilots)
👨💻 Shield AI - Chief Engineer, Autonomy
⚖️ Coreweave - Director, Governance, Risk & Compliance
📰 Everything else in AI today
Microsoft is investigating potential unauthorized data collection from OpenAI’s API by a DeepSeek-linked group, with U.S. AI czar David Sacks also saying there is “substantial evidence” that the company used OpenAI’s models for training.
Microsoft AI CEO Mustafa Suleyman announced that the ‘Think Deeper’ feature is now free for all Copilot users, integrating with OpenAI's o1 reasoning model.
Luma Labs released a new Upscale to 4K feature, enabling video generations on its Dream Machine platform to be upscaled to 4K resolution.
The U.S. Navy banned its members from using DeepSeek for work or personal purposes due to security and ethical concerns.
The ‘Doomsday Clock’ moved to a record 89 seconds to midnight, with scientists citing AI-powered military operations among growing global threats.
Ragon Institute and MIT scientists unveiled MUNIS, an AI tool that accurately IDs viral targets to speed up vaccine design and outperform traditional lab methods.
COMMUNITY
🎥 Join our next live workshop
Join our next workshop this Friday at 4 PM EST to learn about how to automate tasks with OpenAI’s Operator and create your own free personalized, offline assistant with Dr. Alvaro Cintas, The Rundown’s AI professor.
RSVP here. Not a member? Join The Rundown University on a 14-day free trial.
🤝 Share The Rundown, get rewards
We’ll always keep this newsletter 100% free. To support our work, consider sharing The Rundown with your friends, and we’ll send you more free goodies.
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
ChatGPT heads to Washington
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. ChatGPT just got its federal security clearance — and the U.S. government may be getting its most productive civil servant yet.
With OpenAI’s new Gov platform deploying across thousands of agencies, bureaucracy's slow-moving wheels are about to get an efficiency upgrade for the AI age.
Speaking of ChatGPT, our next workshop is this Friday at 4 PM EST. In it, you’ll learn how to automate tasks with OpenAI’s Operator! RSVP here.
In today’s AI rundown:
OpenAI unveils ChatGPT Gov platform
Block releases open-source AI agent platform 'Goose'
Create personalized illustrations for your brand
YuE unlocks open-source AI music generation
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
OPENAI
🏛️ OpenAI unveils ChatGPT Gov platform
Image source: Ideogram / The Rundown
The Rundown: OpenAI just launched ChatGPT Gov, a specialized version of its flagship product designed specifically for U.S. agencies — allowing departments to securely take advantage of AI systems in their own private environment.
The details:
The platform allows agencies to deploy ChatGPT within Azure environments, enabling sensitive data processing and adherence to security protocols.
Gov users can access 4o, and Enterprise features like conversation sharing, custom GPTs, and admin controls for department-wide deployments.
The government has already actively used ChatGPT, with data showing that over 90k employees across 3,500 agencies have generated 18M messages since 2024.
Why it matters: With AI models becoming intertwined in nearly every sector, it's no surprise that the government is adopting the tech internally and needs a more secure deployment for sensitive data and materials. This new, tailored option could significantly spur even broader AI adoption across federal agencies.
TOGETHER WITH RUNNER H
⚡ From text to action
The Rundown: 2025 is the year of AI Agents, and Runner H’s state-of-the-art agent delivers on the promise — transforming text inputs into real-world actions across your organization.
With Runner H, you can easily use natural language to:
Execute intricate tasks like marketing campaigns and code reviews with a single prompt
Automate workflows for instant, real-world business operations
Leverage AI across every aspect of your business
Join the waitlist now to be among the first to test Runner H.
BLOCK
🦢 Block releases open-source AI agent platform 'Goose'
Image source: Block
The Rundown: Twitter founder Jack Dorsey’s fintech company Block just launched Goose, an open-source AI agent framework that allows developers to build and deploy AI assistants across multiple platforms and systems.
The details:
The platform supports any LLM backend, including OpenAI, DeepSeek, and Anthropic, while maintaining data privacy and deployment control.
Goose integrates with Anthropic's MCP and APIs to enable a wide range of tool connections, with the ability to add new integrations mid-session.
Early implementations at Block show Goose automating complex tasks like code migrations, dependency management, and test generation.
The framework was released under the Apache 2.0 license, allowing unrestricted commercial and research use.
Why it matters: The open-source movement is accelerating across AI, and Jack Dorsey’s latest bird-themed release is now bringing it to the agentic layer. Given Block's track record with Square and Cash App, Goose could do what those apps did for AI agents for payments — taking complex tech and making it accessible to everyone.
AI TRAINING
🖌️ Create personalized illustrations for your brand
The Rundown: Freepik lets you create your own branded illustration style by uploading reference images and getting consistent visuals across all your content.
Step-by-step:
Access Freepik AI's Create tab and click the plus icon in the Style section.
Upload your reference images (10-50 images recommended).
Choose your quality setting (Ultra, High, or Medium).
Generate illustrations by selecting your style and writing a prompt.
Pro tip: Check out our exclusive workshop on creating custom styles with Freepik here.
Additionally, Rundown University members enjoy a free monthly premium Freepik plan.
PRESENTED BY CONVEYOR
🛡️ Meet your AI security review sidekick
The Rundown: Conveyor just launched Sue, the first AI Agent for Customer Trust — capable of taking on all the tasks that infosec teams dread like sharing NDA-gated documents, answering security questionnaires and everything in between. Perfect for B2B infosec teams sick of manual work.
Sue can do all of this for you:
Manage and complete customer security requests while communicating across systems and teams
Complete, reject, or escalate questionnaires based on parameters you set
Handle any questionnaire format with no manual tagging — even portals
Personalize your Trust Center using insights from conversation intelligence tools
AI RESEARCH
🎵 YuE unlocks open-source AI music generation
Image source: HKUST
The Rundown: Hong Kong University of Science and Technology researchers just introduced YuE, an open-source AI music generation that transforms lyrics into complete songs, challenging closed commercial platforms like Suno and Udio.
The details:
YuE leverages two specialized models — one for vocals and music and another for production elements — enabling the creation of songs up to 5 minutes in length.
The system supports multiple languages and can handle complex vocal techniques like scatting and mixed-voice performances.
Users can fine-tune their creations with controls for genre, instruments, mood, and vocal characteristics while still maintaining a coherent musical structure.
Why it matters: With commercial AI music platforms like Suno and Udio facing growing legal challenges from record labels, an open-source system could seriously shake up the space. Whether from YuE or another entrant, broader access to AI music tools is coming fast – and the music industry is already reckoning with the changes.
QUICK HITS
🛠️ Trending AI Tools
🎥 Pika 2.1 - New AI video model with enhanced realism, motion, and control
🐋 Janus-Pro - DeepSeek’s new open-source AI image generation model
🤖 Qwen2.5-VL - New family of vision-language models with agentic capabilities to interact with computers and phones
📝 BulletPen - Transform spoken thoughts and rambles into polished writing
💼 AI Job Opportunities
🤝 The Rundown - Partnerships Manager
🔬 Luma AI - Research Scientist
📈 Shield AI - Senior Director of Sales
💼 Findem - Enterprise Account Executive
📰 Everything else in AI today
Alibaba’s Qwen introduced Qwen2.5-Max, a new large-scale MoE model that outperforms DeepSeek-V3, GPT-4o, and Claude 3.5 Sonnet across key benchmarks.
OpenAI CEO Sam Altman chimed in on DeepSeek’s R1 release, calling it an impressive model and saying it’s ‘invigorating’ to have a new competitor.
Figure AI launched a new Center for Advancement of Humanoid Safety to establish industry-wide testing standards and quarterly safety reports for workplace robots.
Ex-OpenAI safety researcher Steven Adler posted on X that he is ‘pretty terrified’ by the pace of AI development, saying no lab currently has a solution for alignment.
Convergence AI launched Proxy, a natural language agent it calls ‘Europe’s answer to OpenAI’s Operator.’
Hugging Face integrated four new serverless inference providers — fal, Replicate, Sambanova, and Together AI — enabling direct model deployment and faster inference.
COMMUNITY
🎥 Join our next live workshop
Join our next workshop this Friday at 4 PM EST to learn about how to automate tasks with OpenAI’s Operator and create your own free personalized, offline assistant with Dr. Alvaro Cintas, The Rundown’s AI professor.
RSVP here. Not a member? Join The Rundown University on a 14-day free trial.
🤝 Share The Rundown, get rewards
We’ll always keep this newsletter 100% free. To support our work, consider sharing The Rundown with your friends, and we’ll send you more free goodies.
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
DeepSeek's ‘Sputnik moment’
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. The AI world is still digesting DeepSeek’s latest R1 breakthrough, but the company is already back with another launch, taking aim at the image generation arena.
With two major releases in a matter of weeks, is tech’s latest disruptor about to upend the AI power balance?
In today’s AI rundown:
DeepSeek launches new AI image model
Qwen launches AI models that control devices
Turn product photos into 3D objects
Meta AI gets a personalization upgrade
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
DEEPSEEK
🐋 DeepSeek launches new AI image model
Image source: DeepSeek
The Rundown: Chinese AI startup DeepSeek just released Janus-Pro, a new open-source multimodal AI model that outperforms major image generation rivals like DALL-E 3 and StabIe Diffusion — coming on the heels of the company’s viral R1 launch.
The details:
The new Janus-Pro model family generates high-quality images from text descriptions, with 1B and 7B parameter models available.
Janus-Pro outperformed DALL-E 3 and Stable Diffusion in key industry benchmarks for image quality and accuracy, such as GenEval and DPG-Bench.
The models were released under an MIT license, allowing developers to freely use and modify the model for commercial projects.
The launch follows DeepSeek's R1 release, which achieved o1-level reasoning capabilities at far lower costs — shaking U.S. markets and the industry.
Why it matters: DeepSeek is the talk of the town, and the effects of R1 are being felt throughout markets as the world digests the reshaping of assumptions around development costs and capabilities. While the current panic may be an overreaction, the Chinese lab has raised questions about the U.S.'s perceived lead in the space.
TOGETHER WITH TURING
📈 Train smarter, scale faster
The Rundown: Turing’s AGI Advancement Hub provides tailored solutions to enhance your language models — helping you optimize performance, reduce costs, and align your models with real-world use cases for everything from multimodality to RLHF.
Explore how to:
Improve reasoning, coding, and multimodal capabilities
Deploy fine-tuned models for enterprise-grade performance
Benchmark effectively with cutting-edge evaluation methods
Visit Turing’s AGI Advancement Hub and start optimizing today.
ALIBABA
🚀 Qwen launches AI models that control devices
Image source: Alibaba
The Rundown: Alibaba’s Qwen team just released Qwen2.5-VL, a new family of vision-language models that can interact with computers and phones, alongside new advanced capabilities in document and video analysis.
The details:
The flagship 72B model outperforms GPT-4o and Claude 3.5 Sonnet on key benchmarks for document parsing and video understanding tasks.
The system can analyze hour-long videos and extract specific moments while processing complex documents like invoices and forms.
A new feature gives the AI agentic control for smartphone apps and computers, with demos including airfare booking, image editing, and code installation.
The smaller 3B and 7B versions are freely available, with the 72B model requiring permission for large-scale commercial uses.
Why it matters: A new ‘operator’ has entered the chat — with Qwen’s computer using vision model coming just a week after OpenAI’s hyped release. Between Qwen and DeepSeek’s massive past week of releases, the gap between open and closed and China vs. U.S. models continues to feel closer than ever before.
AI TRAINING
🎨 Turn product photos into 3D objects
The Rundown: Krea AI’s real-time generation feature lets you create professional product visualizations by combining images, shapes, and AI-assisted enhancements.
Step-by-step:
Access Krea AI's real-time generation feature from the Generate menu.
Upload your product image and position it on the canvas.
Add shapes and painted areas where you want additional elements.
Fine-tune using the AI Strength slider and positioning tools.
Pro tip: Keep AI Strength low for realistic products. Increase it for creative backgrounds and effects.
PRESENTED BY INNOVATING WITH AI
💼 Launch your AI consulting career
The Rundown: Innovating with AI’s new program, AI Consultancy Project, trains AI enthusiasts with the skills and strategies to capitalize on the rapidly growing AI consulting market – which is set to 8x by 2032.
The program offers:
The tools and frameworks to find clients and deliver top-notch services
A 6-month plan to build a 6-figure AI consulting business
Students getting their first AI client in as little as 3 days
META
🧠 Meta AI gets a personalization upgrade
Image source: Meta
The Rundown: Meta just announced the rollout of new AI personalization features that let its assistant remember conversations and tap into user data across Facebook, Instagram, and WhatsApp.
The details:
Meta AI can now remember key details from one-on-one chats, like dietary preferences and interests, to provide more tailored responses.
The assistant will also access users' Facebook locations, Instagram viewing history, and other profile data for personalized recommendations.
The features are launching in the U.S. and Canada across Meta's platforms with no opt-out option, though specific conversation memories can be deleted.
ChatGPT and Gemini have also added ‘memory’ to their assistants, though limited to strictly in-chat and not Meta’s social data and integrations.
Why it matters: Meta has a wealth of social data at its disposal, and tapping into it (similar to its hyper-personalized ads) could give a unique edge with an assistant more in tune with users. However, the lack of an opt-out option feels like a major miss, especially given the company’s complicated history with user data and trust.
QUICK HITS
🛠️ Trending AI Tools
🤖 Qwen2.5-1M - Alibaba’s updated models with 1M token context length
🦙 Llama Stack - Bring GenAI applications to market with unified APIs
📋 OpenAI Canvas - Now available in ChatGPT with o1 and on desktop, with new code rendering capabilities
⚙️ co.dev - Turn ideas into full-stack apps using natural language
💼 AI Job Opportunities
🤖 The Rundown - Writer (Robotics/Tech)
🤝 Waymo - HR Business Partner & Recruiter
🧠 Perplexity AI - LLM Teacher (Contract)
🔬 Luma AI - Senior Research Engineer
📰 Everything else in AI today
LinkedIn co-founder Reid Hoffman announced a $24.6M raise for Manas AI, a new AI-powered drug discovery platform targeting cancer treatments.
DeepSeek’s R1 surged to No.1 on Apple's App Store with 2.6M downloads, but due to reported cyberattacks, it has paused new user sign-ups outside of China.
xAI’s upcoming Grok-3 model briefly appeared to go live for some users, showcasing improved reasoning capabilities — with a full release expected this week.
Pika Labs launched v2.1 of its AI video generation model, which features advanced motion control, realistic physics, and customizable scene elements.
Apple released iOS 18.3, with the new system update turning on Apple Intelligence by default on devices — though AI summaries still remain disabled.
French government AI chatbot Lucie was suspended just days after launch following widespread errors with basic facts, despite funding from President Macron.
COMMUNITY
🎥 Join our next live workshop
Join our next workshop today at 1 PM EST and discover how your musical ideas can become part of your creative toolkit with Sara Adkins, Machine Learning Engineer and Artist in Residence at Suno.
RSVP here. Not a member yet? Join The Rundown University on a 14-day free trial.
🤝 Share The Rundown, get rewards
We’ll always keep this newsletter 100% free. To support our work, consider sharing The Rundown with your friends, and we’ll send you more free goodies.
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
Meta's Manhattan-sized AI play
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. OpenAI’s Stargate Project be hogging the AI infrastructure headlines, but Meta and Zuck just made a Manhattan-sized move of their own.
But with Chinese models achieving similar results on shoestring budgets, are these billion-dollar bets about to reshape AI completely or reveal a massive spending bubble?
Friendly reminder: Our newsletter is still slowly moving to a new email address. Add news@daily.therundown.ai to your contacts to ensure you keep receiving our emails.
In today’s AI rundown:
Zuckerberg announces $65B AI investment plan
Qwen launches model upgrades, 1M token support
How to run DeepSeek R1 offline on your computer
Perplexity AI proposes new TikTok U.S. merger
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
META
🏗️ Zuckerberg announces $65B AI investment plan
Image source: Mark Zuckerberg on Facebook
The Rundown: Meta CEO Mark Zuckerberg just announced a massive $60-65B capital expenditure plan for 2025 focused on AI infrastructure, aiming to position Meta AI as the leading assistant and Llama 4 as the state-of-the-art model in the industry.
The details:
The company plans to deploy roughly 1GW of compute power in 2025, building a datacenter so large it would cover a significant portion of Manhattan.
Meta aims to amass over 1.3M GPUs by year-end, marking one of the largest AI hardware deployments globally.
The investment represents a ~70% jump from 2024's projected spending, with Zuckerberg also predicting that Meta AI will reach 1B users this year.
The news comes on the heels of DeepSeek R1 and OpenAI’s Stargate Project reveal, which will inject $500B into U.S. AI infrastructure projects.
Why it matters: The AI infrastructure race is on, with Meta and OpenAI both pouring unfathomable amounts of capital into the creation of giant new U.S. datacenters. The spending also comes despite DeepSeek’s recent R1 breakthrough — with the Chinese model performing on par with industry leaders despite a fraction of the training costs.
TOGETHER WITH PROJECT IDX
🚀 Boost your development with Cloud + AI
The Rundown: Project IDX offers a complete cloud development environment integrated with Google's Gemini AI — all directly within your browser.
With Project IDX, you get:
A Linux-based virtual machine that you can access anytime, anywhere
Native apps in browser to build and test Flutter/React Native apps with a built-in Android Emulator
Pre-loaded environments to begin coding instantly with popular frameworks
Get up to 5 workspaces and accelerate your development today.
ALIBABA
🤖 Qwen launches model upgrades, 1M token support
Image source: Qwen
The Rundown: Alibaba's Qwen team just unveiled two new open-source models capable of processing up to 1M tokens, alongside significant speed improvements and a revamped Chat interface.
The details:
The new Qwen2.5-1M series includes 7B and 14B parameter models, both supporting 1M token context lengths while maintaining accuracy.
Qwen deploys a custom vLLM-inference framework, delivering up to 7x faster processing than other long-context systems.
In tests, the Qwen-1M models outperformed other long-context models like Llama-3, GLM-4, and GPT-4 across complex long-text tasks.
The release also includes a new Qwen Chat v0.2 upgrade, adding web search, text-to-video generation, and enhanced image capabilities.
Why it matters: Qwen’s open-source 1M models are a hint at whats to come across the industry — with Google’s Gemini (2M) and Flash 2.0 Thinking (1M) already leading the way with massive input capabilities. The context increases, paired with faster processing, open up superhuman levels of data analysis and new complex use cases.
AI TRAINING
💻 How to run DeepSeek R1 offline on your computer
The Rundown: DeepSeek recently released distilled versions of its R1 model, which can match or even outperform GPT-4o and run completely free on your computer.
Step-by-step:
Download LM Studio from their website for your operating system.
Open the app and click the magnifying glass to download a model (recommended: DeepSeek R1 Distill series).
Load your model from the top dropdown menu in the chat interface and start chatting.
Hot tip: These models run completely offline, so you can use them anywhere, even without internet access.
PRESENTED BY PROMPTQL
🧠 RAG with 100% accuracy
The Rundown: PromptQL transforms unpredictable AI into precise agentic intelligence, leveraging agentic planning to deliver a new paradigm of accuracy far beyond the usual 40-70% range of typical RAG and AI agents.
With PromptQL, you can build agents that:
Use dynamic planning to build real-time strategies that adapt to changing data
Have perfect memory to process data without context window limits
Provide full transparency into every decision and output
PERPLEXITY
💼 Perplexity AI proposes new TikTok U.S. merger
Image source: The Rundown
The Rundown: Perplexity AI just revised its merger proposal for TikTok's U.S. operations, introducing a new structure that would give the US government a significant ownership stake in the social media giant.
The details:
The deal would create a new company called ‘NewCo’, combining Perplexity AI and TikTok US, potentially worth as much as $300B after an IPO.
Under the revised plan, the US government could acquire up to 50% ownership — a key sticking point in President Donald Trump’s plan for handling the sale.
TikTok's current owner ByteDance would contribute the U.S. operations but keep the core recommendation algorithm under the proposal.
Other rumored suitors include Elon Musk, Oracle, and Microsoft — with Trump temporarily restoring the app in the U.S. for 75 days to allow for negotiations.
Why it matters: Between its Android mobile assistant and recent API launch, Perplexity is quickly evolving from its original answer engine to an AI powerhouse — but a merger with TikTok U.S. would be by far its biggest move yet. However, with the other tech giants with deeper pockets (like Elon) circling, the plan still feels like a longshot.
QUICK HITS
🛠️ Trending AI Tools
🤖 Operator - OpenAI’s first agent for in-browser tasks
📱 Perplexity Assistant - Android agentic assistant capable of controlling phone apps and performing complex tasks
📚 Citations - Allow Claude to ground answers in source documents and provide references
📣 WePost - Generate on-brand marketing with ease
💼 AI Job Opportunities
📣 The Rundown - Marketing/Media Buyer
🤝 Lindy AI - Head of Community
📊 Weights & Biases - FP&A Associate, R&D (US)
🧱 Databricks - Solutions Architect
📰 Everything else in AI today
ElevenLabs is reportedly raising a $250M Series C at a $3B+ valuation, with demand surging for its AI voice synthesis and dubbing technology.
Anthropic CEO Dario Amodei predicted that AI could enable humans to live 2x longer by 2030, with the tech compressing a century of research progress into 5-10 years.
xAI is reportedly developing a voice interface for its Grok iOS app with both proprietary and ElevenLabs voice options, also capable of leveraging real-time data.
OpenAI expanded Canvas functionality in ChatGPT with new rendering capabilities and o1 model support, also rolling out desktop app access across all subscription tiers.
Legendary musician Paul McCartney spoke out against the UK’s proposed AI copyright law changes, warning they could “rip off” musicians without compensation.
OpenAI CEO Sam Altman said that advances in AI will eventually require changing the social contract, with “the whole structure of society up for debate and reconfiguration.”
COMMUNITY
🎥 Join our next live workshop
Join our next workshop on Tuesday at 1 PM EST and learn how to turn ideas into original music with our next expert guest: Sara Adkins, Machine Learning Engineer and Artist in Residence at Suno.
RSVP here. Not a member yet? Join The Rundown University on a 14-day free trial.
🤝 Share The Rundown, get rewards
We’ll always keep this newsletter 100% free. To support our work, consider sharing The Rundown with your friends, and we’ll send you more free goodies.
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
OpenAI's first AI agent arrives
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. 2025 has already been widely declared the year of AI agents, and OpenAI just officially joined the party.
The startup’s ‘Operator’ release takes us into a new realm with mainstream AI assistants that can navigate the internet and take actions on their own — our interactions with chatbots may never be the same.
Exclusive: I got early access to test Operator, which did not disappoint. Check out my thread of demos here.
In today’s AI rundown:
OpenAI unveils its first autonomous web agent
Perplexity debuts new AI mobile assistant
How to prompt o1 models better
‘Humanity’s Last Exam’ scales up AI benchmark
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
OPENAI
🤖 OpenAI unveils its first autonomous web agent
Image source: OpenAI
The Rundown: OpenAI just launched Operator, an AI agent that can independently navigate web browsers to complete everyday tasks — marking the company's first major step into autonomous AI assistants.
The details:
Operator uses a new Computer-Using Agent model that combines 4o's vision capabilities with advanced reasoning to interact naturally with websites.
OpenAI demoed the feature during a live stream, showcasing tasks like booking reservations, grocery ordering, and buying tickets to sporting events.
OpenAI has partnered with major platforms like DoorDash, Instacart, and Uber to ensure the agent works seamlessly while respecting platform guidelines.
Built-in safety features include user approval for purchases, automated threat detection, and "takeover mode" for sensitive info like passwords and payments.
The research preview is currently limited to U.S. Pro users, with plans to expand to Plus, Team, and Enterprise after more safety and reliability testing.
Why it matters: While we’ve seen agentic systems popping up more frequently, OpenAI’s long-awaited move is a major step towards broadly changing the entire mindset of how we interact with AI. While there may be rough edges at first, Operator feels like the official beginning of a brand new agentic era.
TOGETHER WITH WORKOS
🔐 Protect your AI apps from bad actors
The Rundown: WorkOS Radar is a security solution that shields your AI platform from fake signups, throwaway emails, and brute force attempts — all powered by advanced device fingerprinting and real-time detection.
With WorkOS Radar, you can:
Rapidly detect and challenge unfamiliar and suspicious devices in real time
Stop free-tier abuse and fraudulent behavior with advanced detection
Customize threat responses to fit your app’s exact security needs
PERPLEXITY
📱 Perplexity debuts new AI mobile assistant
Image source: Perplexity
The Rundown: Perplexity just unveiled Perplexity Assistant, a free, agent-like tool for Android that can control phone apps and perform complex tasks with multimodal and voice capabilities — directly challenging voice assistants like Google’s Gemini and Siri.
The details:
The new assistant integrates with popular apps like Uber and OpenTable to perform actions directly through voice commands or gesture controls.
It maintains context throughout interactions, allowing users to progress from research to action — like finding restaurants and booking a table.
The system supports multimodal interactions through voice and camera, enabling users to obtain information about their surroundings or view screen content.
Users can replace Google's default assistant with Perplexity's solution at no cost, with the feature only available on Android for now.
Why it matters: Operator isn’t the only agent in town today, with Perplexity evolving its platform from a search/answer engine to a full-blown digital assistant. The assistant space could become a new battleground for AI firms, not just tech giants — and this Perplexity launch looks a lot like what Apple’s ‘upgraded’ Siri should actually be.
AI TRAINING
🤖 How to prompt o1 models better
The Rundown: Using delimiters and structured prompts significantly improves AI model outputs by providing clear instructions and relevant context.
Step-by-step:
Structure prompts using XML tags (<goal>, <context>, <format>).
Fill each section with relevant, specific information and define clear parameters for your desired output.
Test, ask follow-up questions, and optimize your results.
Pro tip: Save successful prompt structures as templates for consistent results across similar tasks. The Rundown University members can access our full workshop on effectively using ChatGPT-o1 to get the best results here.
PRESENTED BY INNOVATING WITH AI
🤝 Turn AI passion into a consulting career
The Rundown: Innovating with AI's new program, AI Consultancy Project, transforms AI enthusiasts into professional consultants — tapping into a market projected to reach $54.7B by 2032.
The 6-month program delivers:
Proven frameworks for client acquisition and service delivery
A step-by-step path to six-figure consulting income
Students who land their first AI client in as little as 3 days
Click here to request early access to The AI Consultancy Project.
SCALE AI & THE CENTER FOR AI SAFETY
🧐 ‘Humanity’s Last Exam’ scales up AI benchmark
Image source: Humanity’s Last Exam
The Rundown: The Center for AI Safety and Scale AI just introduced "Humanity's Last Exam," a new AI benchmark designed to be the final frontier for testing an LLM’s academic knowledge — as current AI systems become too strong for existing tests.
The details:
The benchmark consists of 3,000 expert-crafted questions across 100+ subjects, with contributors from over 500 institutions in 50 countries.
Current leading AI models show surprisingly low performance on HLE, with even top systems scoring under 10% accuracy.
Questions are in either exact-match or multiple-choice format, with 10% of the challenges incorporating multimodal analysis of text and images.
A $500k prize pool incentivizes high-quality submissions, with top questions earning $5,000 each and co-authorship opportunities for contributors.
Why it matters: With top models routinely scoring above 90% on many of today’s key benchmarks, tests like HLE are an important way to continue scaling the ability to measure increasingly advancing AI systems. However, given the rate of progress, it likely won’t be long before we see some impressive results on these benchmarks.
QUICK HITS
🛠️ Trending AI Tools
🤖 Gemini 2.0 Flash Thinking Exp - Google’s powerful new reasoning model
⚙️ Trae - Adaptive AI IDE that helps you ship faster
💻 UI-TARS - Control your computer using natural language
🪄 Spell by Spline - Generate full 3D scenes or worlds from a single image
💼 AI Job Opportunities
🎯 The Rundown - Marketing/Media Buyer
👋 Lindy AI - Head of Community
📊 Snorkel - Data Annotator (Statistics Expert Contributor)
🗣️ UiPath - ASR Manager
📰 Everything else in AI today
Anthropic launched Citations, a new feature in the Claude API that enables automated source attribution and verification in responses for increased accuracy.
Google’s Imagen 3.0 debuted at No. 1 in the LM Text-to-Image Arena, giving the tech giant the top spots on image and LLM leaderboards.
ByteDance is planning a $20B investment in AI infrastructure in 2025, half of which will be allocated to international data centers and partnerships with chip suppliers.
OpenAI CEO Sam Altman revealed that the upcoming o3-mini model upgrade will be available in the free tier of ChatGPT, with usage upgrades for plus users.
Hugging Face unveiled SmolVLM 256M and 500M, hailed as the world’s smallest vision language models that maintain competitive performance against larger rivals.
LinkedIn is facing a new class-action lawsuit, alleging that the company used the private messages of premium subscribers to train AI models.
COMMUNITY
🎥 Join our next live workshop
Join our next workshop today at 4 PM EST to learn about how to use DeepSeek-R1 as a powerful, cost-effective alternative for your AI projects with Dr. Alvaro Cintas, The Rundown’s AI professor.
RSVP here. Not a member? Join The Rundown University on a 14-day free trial.
🤝 Share The Rundown, get rewards
We’ll always keep this newsletter 100% free. To support our work, consider sharing The Rundown with your friends, and we’ll send you more free goodies.
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
Exclusive: Cisco launches AI Defense
Good morning, AI enthusiasts. In case you missed it, Cisco just announced AI Defense — a new solution that aims to become the global standard for AI security.
In a future where workforces will include AI workers — including apps, agents, robots and humanoids, understanding AI security is essential for every knowledge worker and business.
So we partnered with Cisco and Jeetu Patel, Executive Vice President and CPO at Cisco, for an exclusive Q&A on how AI Defense is tackling this challenge.
Free event: Join Cisco’s Virtual AI Summit to hear from the most influential figures in AI, including the COO of OpenAI, Scale AI, Dr. Fei-Fei Li, and more.
In today’s AI rundown:
Cisco’s AI Defense: A new era for enterprise security
AI Defense's two-fold data strategy for sensitive info
Protection at scale: Network-level security integration
The future of multi-model, multi-cloud AI security
Why AI-forward is the only way forward
EXCLUSIVE Q&A JEETU PATEL
ANNOUNCEMENT
🛡️ Cisco’s AI Defense: A new era of AI security
Image credits: Kiki Wu / The Rundown
The Rundown: Cisco announced AI Defense, a new security solution designed to protect AI systems in a future where workforces will expand to include AI workers — including apps, agents, robots, and humanoids.
Cheung: "AI security is dominating headlines right now. But for our readers who may be new to the topic, can you explain in simple terms what AI Defense is and why this announcement is a big deal?"
Patel: "The reality is that there will be two types of companies in the future: those who are leading with AI and those that are irrelevant. Every company will be using - if not developing - thousands of AI applications, and the rapid pace of AI innovation is outperforming major security and protection concerns."
Patel added: "We developed AI Defense to protect both the development and use of AI applications. Overall, AI Defense safeguards against the misuse of AI tools, data leakage and increasingly sophisticated threats. We are taking a radical approach to address the challenges that existing security solutions are not equipped to handle."
Why it matters: Enterprises are racing to adopt AI, and due to the rapid pace of the space, traditional security solutions are falling behind. Cisco aims to make AI Defense the global standard for AI security in the increasingly AI-powered world.
STRATEGY
🧠 AI Defense's two-fold data strategy
Image credits: Kiki Wu / The Rundown
The Rundown: Cisco’s AI Defense combines two protections: monitoring third-party AI app usage and protecting sensitive data in custom AI development — addressing the growing risks of data leakage and unauthorized access.
Cheung: "A lot of enterprises are now training their own AI models on sensitive data. What specific protections does AI Defense offer that weren't possible before?"
Patel: "Enterprises are training their models on proprietary data and connecting backend data sources to create more useful AI applications. This expands the risk surface tremendously; any data connected to your AI application is now subject to leakage, poisoning, or exfiltration."
Patel added: "AI Defense takes a two-fold approach. First, it gives security teams visibility into third-party AI application usage, letting them enforce policies to limit data sharing and restrict access to high-risk solutions."
Patel added: "For enterprises developing and deploying their own AI applications, AI Defense examines user inputs and model outputs to block sensitive data leakage in real-time. This covers scenarios like an adversary attempting to extract PII or an AI application inadvertently exposing snippets of proprietary source code, for example."
Why it matters: As enterprises adopt both third-party and custom AI solutions, the risk of data leakage grows exponentially. AI Defense's two-fold approach allows organizations to innovate faster with AI while keeping their sensitive data secure. Patel goes into more depth on this strategy in his blog post.
PROTECTION
🔐 Protection at scale: How AI Defense works
Image credits: Kiki Wu / The Rundown
The Rundown: Unlike traditional security solutions, Cisco’s AI Defense is integrated directly into network infrastructure, providing consistent controls and protection at scale across all AI apps.
Cheung: "You mentioned AI Defense is 'fused into the fabric of the network' - this sounds fascinating. Can you break down what this means in practical terms and why it matters?"
Patel: "Simply put, this means that AI Defense is deeply integrated into the overall network infrastructure. By doing security at the network layer, AI Defense provides consistent controls for AI applications across the business."
Patel added: "This removes a massive burden for your AI developers. They would otherwise spend tremendous time and resources trying to secure individual applications in accordance with broader security policies."
Patel added: "This approach also leverages Cisco's existing network mesh to provide unparalleled visibility and control over AI traffic on our customers' networks. As security fuses further into the network and cloud with Hypershield, the ability of AI Defense to identify and secure east-west AI-related traffic will grow exponentially."
Why it matters: By integrating security at the network level, Cisco is changing how enterprises protect AI applications. With thousands of distributed enforcement points across clouds and servers, this approach improves security while freeing developers from application-by-application implementation.
INTO THE FUTURE
🚀 The future of AI security
Image credits: Kiki Wu / The Rundown
The Rundown: As enterprises move into a multi-model, multi-cloud world, Cisco is preparing AI Defense to tackle the growing complexity of AI security — where vulnerabilities can exist at both model and application levels.
Cheung: "We've seen AI evolve incredibly fast over the past couple years. How do you see AI Defense evolving to handle increasingly sophisticated threats, especially as AI agents become more prevalent?"
Patel: "You're right, we're already in a multi-model, multi-cloud world, and 2025 will be the year where companies start to deploy AI applications in a big way. The complexity is growing every day, and so is the risk."
Patel added: "This 'complexity' is a hard computer science problem to solve, since AI apps are built and operate differently. Ownership is fractured, meaning vulnerabilities can be at model or app level, and responsibility lives with different people – such as security practitioners, developers, users, vendors, and so on."
Why it matters: AI applications are becoming increasingly complex, and security challenges will continue to evolve. Cisco's AI Defense aims to take out the complexity by unifying security across models, applications, and stakeholders.
TRANSFORMATION
📈 Why AI-forward is the only way forward
Image credits: Kiki Wu / The Rundown
The Rundown: Jeetu Patel believes that companies who are not AI-forward will no longer be relevant in the future. With AI Defense, Cisco aims to remove the biggest barrier to adoption at the enterprise level — safety and security concerns.
Cheung: "As we enter this new era of AI adoption, how do you think AI Defense will change the way enterprises build and deploy AI applications?"
Patel: "Companies who are not AI-forward will no longer be relevant. Every application will be an AI application, and every company will be using hundreds—if not thousands—of these apps every day."
Patel added: "Our Cisco Readiness Index confirmed that one of the biggest roadblocks to AI adoption is concerns around safety and security. We built Cisco AI Defense specifically to address these concerns, and to allow enterprises to develop, deploy, and use AI applications with confidence."
Why it matters: Security concerns are the biggest barrier preventing enterprises from becoming AI-forward. Cisco's AI Defense removes this roadblock, allowing companies to confidently build and deploy more AI applications — ultimately benefiting the end users who can rely on these tools for the future of AI-enabled work.
GO DEEPER
AI SUMMIT
🤩 Free event: Join Cisco’s Virtual AI Summit
Join Cisco’s AI Summit on January 22, 2025, for an exclusive virtual broadcast featuring groundbreaking sessions with:
OpenAI COO Brad Lightcap, to explore how AI is transforming industries like financial services, healthcare, and telecom
Scale AI CEO Alexandr Wang on the current realities of leveraging data to build better models
Dr. Fei-Fei Li on how AI is blurring the boundaries between virtual and physical, enabling unprecedented customization, efficiency, and engagement
Plus more from the CEOs of Goldman Sachs, Cohere, Box, Glean, and Groq.
Register for free and witness some of the greatest minds coming together to shape the future of AI.
No matching search results
Try using different keywords, double-check your spelling, or explore related categories.
Stay Ahead on AI.
Join 2,000,000+ readers getting bite-size AI news updates straight to their inbox every morning with The Rundown AI newsletter. It's 100% free.