- AI Fire
- Posts
- 🚜 Farmer Uses Codex to Run Crops
In partnership with
Anthropic dropped a two-part policy framework, backed by a preview model so powerful it discovered zero-day vulnerabilities across every single major browser.
What's on FIRE 🔥
IN PARTNERSHIP WITH SECTION
If you’re like a lot of other leaders today, you probably don’t feel ready to deploy AI agents in your company.
On 6/18, Section is giving away the playbook, for free. You’ll leave with the guide to getting your agentic strategy started and what you actually need to deploy agents ASAP on your team.
AI INSIGHTS
AI is moving at a terrifying, exponential speed, while the policymaking world is practically built for a slower era. To close that gap, Anthropic dropped a two-part policy framework.
To prove this, it dropped a preview version of their next-gen model, Claude Mythos Preview, just discovered thousands of high-severity software vulnerabilities, including zero-days in every single major operating browser:
Who this applies to: These strict rules would only hit true "frontier" developers, companies pulling in $500M+ in AI revenue, or spending $1B+ on AI.
The framework specifically targets 4 scenarios: AI lowering the barrier to creating biological weapons, massive automated cyberattacks on critical infrastructure, models completely losing control and acting outside developer intent, and AI accelerating its own R&D loop into a runaway spiral.
Frontier labs would be legally forced to run rigorous internal testing, publish risk reports, and hand their models over to qualified independent evaluators.
Anthropic warns that Congress shouldn't block powerful state-level AI laws (like those in California or New York) unless the federal law is just as strict.
Beyond just stopping bad models, the framework outlines a "societal resilience" plan to start mandatory gene synthesis screening to prevent bioweapons, and rapidly patching the aging software that runs our critical power grids and hospitals.
PRESENTED BY HUBSPOT
Want to get the most out of ChatGPT?
ChatGPT is a superpower if you know how to use it correctly.
Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.
Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.
AI SOURCES FROM AI FIRE
1. Free Analysis: Claude Fable 5: Unlock the Model Anthropic was Afraid to Release. What can Claude Fable 5 really do? We uncovered hidden capabilities Anthropic almost kept secret. You won’t believe what it handles.
2. [How To AI in Real Estate] Lesson 5: Building Your Lead Intake System in Systeme. Turn your client journey map into a structured CRM workflow, automate internal notifications with Zapier, and use Claude to review leads before follow-up.
3. Video: Everything You Must Know Before Using Gemini Omni. See how to use Google Flow and Gemini Omni to control virtual cameras, stack effects, test AI avatars, and build a full AI video production workflow with basic prompts.
CHATGPT CODEX FULL COURSE
I know you've seen it. Your feed, my feed, everyone going crazy comparing Claude Code vs ChatGPT Codex. Creators are going viral off it, devs are picking sides, and honestly? The hype is real.
But I kept noticing that nobody's actually sitting down and showing beginners how to use Codex. Like, from zero.
Now, last time I gave you a full course on Claude Code, and you guys loved it. So this time, I thought: let's do the other side. Let's go deep on Codex.
I'm going to walk you through everything, and I mean everything you need to get started with Codex in 2026. You'll actually be able to use it. 🚀
TODAY IN AI
AI HIGHLIGHTS
📝 Based on the leaked system prompt, here's how to prompt Fable 5 properly. While unofficial (yet insightful), it shows Claude is tool-heavy, safe, and info-aware.
🤯 Less than a day in, Claude Fable 5 is blowing minds. It aced 3 of the internet’s trickiest questions designed to fool AI (that 'How many Rs'? test). Folks are sharing wild ideas, here’re 10 crazy use cases you’ll want to see.
🎬 HeyGen is an official Claude connector. You can create videos from chat, with 25+ skills for editing scenes, motion, and voice, rendering directly to multiple formats.
🤔 OpenAI is considering major token price cuts, potentially ahead of anticipated Anthropic’s moves. You could see cheaper access soon, but nothing is finalized yet.
🎬 Luma AI released Ray3.2, letting you direct video with text prompts. New features include 16 keyframes, 8 face tracks, 20-second clips at 1080p, a full API. Try it here.
🚜 A Japanese farmer with no engineering background just built his own greenhouse automation, satellite crop tracker, and farm bot, using Codex. Here’s how he did it.
💵 Perplexity is sticking to 2028 for its IPO, even with Anthropic and OpenAI gearing up. Will the mega-IPOs affect their plans? Probably, but not enough to delay them.
💰 Big AI Fundraising: Standard Bots raised $200M at a $1B valuation, claiming its AI robots can boost U.S. manufacturing. Worker count dropped from 20M in 1979 to 13M today, robots aim to create jobs.
Simply upload a PDF, Word document, paste a URL, anything. SlidesPilot understands the content, organizes the key points, creates a structured, editable presentation with layouts and visual elements.
YouTube videos to presentation decks
Export as PowerPoint (PPTX), Google Slides, PDF, or PNG
Fully editable after generation
AI-generated charts, diagrams, and visuals
Whether you're creating client presentations, pitch decks, reports, training materials, or class lectures, SlidesPilot can help you create slides up to 10X faster.
NEW EMPOWERED AI TOOLS
📣 Publora is a publishing API for 10 social platforms, with 18 tools gives AI agents like Claude and Cursor a full engagement loop.
🧠 TypingMind brings you all the best models across 18 providers in one workspace without having to pay for subscriptions to each one.
🔍 Spotlight reads your Claude Code and Codex sessions and shows you what your agents actually did. Free to use.
🎥 Screen Charm records your screen & instantly turns raw footage into polished videos with smart zoom effects, and a built-in editor.
AI BREAKTHROUGH
Perplexity and Harvard Business School dropped a fascinating study comparing Perplexity's "Computer" agent platform against regular Search, and it totally shifts how we should look at knowledge work.
They ran 10,000 identical queries through both paths to measure the difference in output, time saved, and task complexity. Here is what the data showed:
Regular search feels quick. But search leaves the heavy execution to you. When you factor in the human effort, the Search workflow took an estimated 269 minutes. The agent wrapped the whole workflow up in 36 minutes.
Half of the tasks users handed to the agent involved building something entirely new. That is double the creation rate we see on regular search.
The number of tasks that fell completely outside the user's actual field of expertise jumped up 9 points to 59%.
Users trusted the agent with cognitively heavy work, asking it to generate code, draft docs, and build visuals across multiple disciplines.
The real unlock here is human ambition. When you remove the friction of the grunt work, people naturally start aiming higher. Perplexity's data proves that interacting with agents gives us the confidence to handle complex, creative projects we normally wouldn't even attempt.
We read your emails, comments, and poll replies daily
How would you rate today’s newsletter?Your feedback helps us create the best newsletter possible |
Hit reply and say Hello – we'd love to hear from you!
Like what you're reading? Forward it to friends, and they can sign up here.
Cheers,
The AI Fire Team
Reply