Did you know that we have LinkedIn, Instagram and X accounts
that you can follow?
Hi {{first_name|everyone}},
At the UN Security Council, some of the people building the most powerful AI systems were the ones asking for more guardrails, though they didn't agree on what those should look like. After that, we've got a few handy new tools, some interesting discussion from the community, and a song that got remixed 6 different ways.
Today's sponsor is Pitch, and their MCP lets you build polished decks from your existing work right inside Claude, so let's get into it.
🤝 FROM OUR FRIENDS
Ask for a deck. Get a deck.
Claude already talks to your CRM, your notes, and your docs.
Now it talks to Pitch too.
With Pitch MCP, connect your workspace once and build decks through plain conversation. Feed it a call transcript and a CRM record, and get a client‑ready proposal back as a shareable link, in the same chat where you asked for it.
Want the output locked to spec? Create from templates with variables for things like customer names, dates, or pain points instead, so every deck matches your brand.
It doesn’t stop at creation: browse teamspaces and folders, pull engagement analytics into the thread, or lift content from an old deck into a new one.
Automating at scale? Pitch API puts the same logic behind Zapier, your CRM’s workflow builder, or your own tools.
🤿 DEEP DIVE
When AI Outruns the Rules
The shared thread was human control, even when the fixes differed.
The slightly strange thing about this Security Council meeting was who was asking for more guardrails.
Not just governments. Some of the people building the most powerful AI systems were saying it too.
They definitely didn’t agree on what those guardrails should look like. But the same basic problem kept showing up. AI is getting more capable and more autonomous, and the systems meant to keep it under control have to keep up.
Yoshua Bengio gave the starkest version of that argument. He pointed to recent cases where AI agents reportedly bypassed controls, coordinated cyberattacks and tried to hide what they were doing. His concern wasn’t only that people could misuse powerful AI. It was that the systems themselves could become difficult to control.
And that distinction matters.
There’s the familiar problem of someone using AI for cyberattacks, biological weapons or disinformation. Then there’s the newer problem discussed throughout the meeting, where increasingly autonomous systems behave in ways their developers didn’t intend.
Bengio wants independent safety evaluations, incident reporting, licensing for frontier AI and international agreements.
Sam Altman came at it differently, but landed on some of the same territory. He said companies shouldn’t train systems they can’t make a very strong case will remain under human control. He also called for common standards around capabilities, risk testing, safeguards and incident reporting.
Dario Amodei proposed external evaluators, verification systems and global standards for testing models for both misuse and loss-of-control risks.
Then Hugging Face’s Clément Delangue added another wrinkle. His concern was that safety rules themselves can create an imbalance if defenders are blocked from powerful tools while attackers find ways around the restrictions. His answer leaned more toward transparency, incident disclosure and open-source access.
So this wasn’t a room full of people agreeing on one AI rulebook.
What they were circling around was something more basic. A technology that crosses borders, can be used by governments and non-state actors, and may become harder for its own developers to understand was repeatedly framed as too broad to leave to individual companies alone.
That’s where the UN comes in.
Across the meeting, speakers kept returning to the same kinds of mechanisms. Common standards. Shared incident reporting. Independent evaluation. More transparency. More international cooperation.
But there was still a real disagreement over how far that cooperation should go.
The US representative supported international dialogue but rejected new global governance structures, arguing that countries should retain national sovereignty over how they regulate and deploy AI. Russia also pushed back on treating the Security Council as the main venue, saying much of the work belongs in more specialized forums.
So the meeting didn’t settle the question of who should govern frontier AI.
It clarified the argument.
The technology is still moving.
The question now is whether the rules can move with it.
The quick take
Bengio framed frontier AI around 3 risks including misuse, concentrated power and loss of human control
Altman and Amodei both backed stronger evaluations, incident reporting and international standards
Hugging Face argued that transparency and open-source access matter because safety measures can create asymmetries between attackers and defenders
Governments broadly agreed on the need for human control, but differed over how much authority international institutions should have
UN efforts already include the Global Dialogue on AI governance and the Independent International Scientific Panel on AI
🔥 WHAT’S THE BUZZ
What people are building, sharing, and talking about
I usually think of AI in a writing interface as the thing doing the writing. This is a little different. It just sits there judging what you’re making as you make it. That opens up some interesting interface ideas.
⚒ TOOL SNAPSHOTS
Futuristic tools within AI, no-code, and productivity
🧠 Hemory
Turns the conversations you hear all day into memory you can search later.
Why it’s useful: Handy if important context tends to live in conversations, especially since you can bring that memory into tools like Claude, Codex, and Cursor through MCP.
🎥 Eclatira
Build agents that can talk, see, and act while a conversation is happening.
Why it’s useful: Gives developers one layer for voice, live vision, and tool calls, instead of stitching together separate systems for each part.
🔍 Cuey
See where other AI models disagree before trusting one answer.
Why it’s useful: Useful for research, advice, or judgment-heavy questions where a confident response can still miss something and checking several models manually gets annoying.
🎁REFERRAL REWARDS
If TIW has been useful, invite one friend or coworker who’d actually use AI/automation at work.
1 referral → get the TIW No-Code Workflows PDF (25 plug-and-play workflows)
10 referrals → get a full course free (your choice)
Your referral link: {{rp_refer_url}}
Tip: send it to a friend who’s AI-curious, a coworker who wants to move faster, or anyone trying to automate repetitive tasks.
🤖 AI MADE THIS
Interesting and inspiring creations made with AI
Suno V6 turns one song into 6 completely different piano versions
What happens if you keep the same AI-generated song but swap out just one instrument? Papiron tested Suno V6 by creating six versions with different piano and guitar sounds, making the model's new control over individual musical elements surprisingly easy to hear.
AI used
Suno V6, specifically its Cover feature. The creator also tested re-covering an older V5.5 song with V6.
ℹ️ ABOUT US
The Intelligent Worker helps you to be more productive at work with AI, automation, no-code, and other technologies.
We like real, practical, and tangible use-cases and hate hand-wavy, theoretical, and abstract concepts that don’t drive real-world outcomes.
Our mission is to empower individuals, boost their productivity, and future-proof their careers.
We read all your comments - please provide your feedback!
Did you like today's email?
What more do you want to see in this newsletter?
