Freedom OS Now Live: Learn exactly what AI Business Model to start based on YOUR situation. Use code EARLYBIRDOS for 50% off (expires midnight tonight, UK time).
The skinny on what's happening in AI - straight from the previous live session:
Highlights:
🚨 Chinese Hackers Used Claude Code for Mass Cyber Espionage
Anthropic discovered Chinese state-sponsored hackers used Claude Code to infiltrate 30+ organisations including banks, tech firms, chemical manufacturers, government bodies. 80-90% of work done by AI, minimal human intervention. Uh-oh…
Beautiful explainer graphic!
Kyle's take: The AI wars have begun? This signals AI-driven espionage is feasible at scale - but let's be honest, it's probably been happening for ages and we just found out.
Anthropic came forward partly for damage control (their tool was used) but also bizarre positive marketing: "Our tools are so good, Chinese state-backed hackers choose them!" Great testimonial in a weird way.
They published a nice flowchart showing exactly how the attack worked - phase 1: target selection, phase 2: vulnerability scanning, phase 3: iterative exploitation, phase 4: credential theft. Thanks for the manual, Anthropic!
The real worry isn't this specific attack - it's the gap widening between AI users and non-users. UK's particularly screwed because we cap civil service cybersecurity salaries at £100k while top talent goes to America for £2M++. Our Department of Defence can be peeled open like a tin while we're arguing about civil service pay bands…
💰 ByteDance Launches $1.30/Month AI Coding Agent (Good Luck Finding It)
TikTok's parent company ByteDance unveiled "Volcano Engine" - AI coding agent for 10 yuan ($1.30) first month, then $4-5 monthly. Includes access to Claude Code Pro.
Kyle's take: Everyone's hyping this on Twitter but I bet none of them have actually used it. Why? The entire site is in Chinese. You need to navigate to volcengine.com, find the right page, scroll through Chinese text, sign up with a phone number.
I found it - took me ages. And I speak Mandarin.
Sure this is China saying "we can undercut Western AI tools by 95%" but they're not actually trying to capture Western market. They don’t need to. Hence zero English interface.
🦄 Cursor Raises $2.3B at $30B Valuation on $1B Revenue
AI coding IDE Cursor raised Series D from Accel, Andreessen Horowitz, Nvidia, Google. Now produces "more code than any other agent in the world."
Kyle's take: $30 billion valuation on $1 billion revenue - bubble or future? Here's what people may not realise: most of that revenue goes straight through to model providers (OpenAI, Anthropic, Google). Cursor's basically a shopfront charging us then paying most of it to Claude and GPT.
They're a middleman with a nice interface. Apparently the margin on model usage is 20%.
Great product, I use it daily, but when ByteDance offers similar for $1.30, you see the problem….
📢 Freedom OS Launched this weekend: AI Business Assessment Tool
Over the weekend I launched the Freedom OS - 7-step assessment tool that analyses your situation and recommends which AI business model suits you. Early bird: $47 (goes back to $97 TODAY, Monday at midnight).
Kyle's take: I keep getting asked "what AI business should I start?" - impossible to answer without context. This tool assesses: threat level to your job, income needs for freedom, AI focus areas, current AI skill level (1-8 scale), then recommends specific business models. Level 1 is "I chat with GPT sometimes," Level 8 is "I read research papers and fine-tune models." The tool gives you an action plan based on YOUR situation, not generic "start a faceless YouTube channel" nonsense.
Kyle's response: Very fast, incredibly optimised, but hasn't replaced Claude Code for me. Composer rockets through tasks while Claude Code makes me wait - but Claude's results are better. For m at least! If I was paying for Claude's capacity instead of having free 20x max account from Anthropic, I'd probably use Composer. Also hearing GPT 5.1 in Codex is good but haven't tested properly.
Kyle's response: DON'T mass produce AI blog spam - Google will slap you down eventually. Start with human content: I talk for an hour (~7,500 words of content), transcript becomes newsletter, newsletter becomes SEO articles. Each story gets expanded with original transcript. Human source material makes it less AI-slop and speaking generates 10x more content than typing. Those "50 blog posts daily" services are scams.
Kyle's response: Only for huge context window tasks. Tried Imagen 3 yesterday for graphics - complete failure. Asked for changes, it just gave same image repeatedly. Went back to DALL-E which worked perfectly. For me, Claude handles most things, ChatGPT fills gaps, Gemini sits unused.
(Note: we’re probably seeing Gemini 3.0 this week - so this may all change!)
Want the full unfiltered discussion? Join me tomorrow for the daily AI news live stream where we dig into the stories and you can ask questions directly.
Streaming on YouTube (with screen share) and TikTok (follow and turn on notifications for Live Notification).
