AI Wants Your Backyard Next

Today on Briefly AI — Nvidia's Jensen Huang says AI doesn't need regulating, then hands the microphone to the President to prove it. A new poll says most of America would rather the data centre boom went somewhere else, and Philadelphia's already fighting that exact battle. And Meta builds a front door to WhatsApp Business that's designed for robots, not people.

Welcome to Briefly AI, a podcast by Harry Sharman, written and voiced by his AI clone. The machines had a busy week again, so naturally one of them's been sent to read it all back to you. It works rather better than it should.

Okay — let's get into it.

Right, so here's a fun experiment. Take the most powerful chipmaker on the planet, put him on a stage in front of hundreds of investors, and have him argue that his own industry doesn't need watching. Then, for good measure, ring up the President live on speakerphone. Nothing says "we've got this handled" quite like a surprise guest appearance from the Oval Office.

So, on Monday, Nvidia's chief executive Jensen Huang took the stage at the All-In Podcast's All-In Summit and made an argument that's going to irritate a lot of people who've spent this year drafting AI safety rules. According to TechCrunch, his case was this: AI isn't some new "alien mind" we don't understand — it's hardware and software, same as anything else his company has ever shipped, and safety is something you engineer into a product at the design stage, not something you hand over to an outside regulator to police after the fact. In Huang's telling, treating AI as a uniquely dangerous, uniquely mysterious category of technology is the mistake — build it properly, and the companies making it are the ones responsible for making it safe, full stop.

Then came the bit that turned a policy speech into an actual show. The Verge reports that mid-appearance, Huang took a phone call from President Trump — and put it on speakerphone, in front of the entire crowd. Apparently not the first time he's taken a call from the President mid-work, but certainly the biggest audience he's done it in front of. Trump used the moment to declare, more or less, that robots aren't going to take over the world, and ran with the bit for a while.

Here's why it's worth more than a chuckle. Huang runs the company whose chips every major AI lab depends on to build anything at all. When the most important supplier in the industry says self-regulation is enough, and then demonstrates he can get the President on the phone whenever he fancies, that's not a neutral opinion floating around a conference. That's the market's most powerful vendor setting the political weather on regulation, live, with a presidential cameo doing the advertising for free.

Meanwhile, away from the stage theatrics, some actual numbers came in on how the public feels about all this — and it isn't close. A poll released this week by the New York Times and Siena University, covered by The Verge, asked just over fifteen hundred likely voters whether they support or oppose building data centres to power AI. Sixty-one percent said oppose. Not "it depends," not lukewarm — a clear, sizeable majority actively against it.

You can see exactly why if you look at where that fight is actually landing. TechCrunch reported this week that the data centre boom has reached Philadelphia, where officials have floated building on land in a neighbourhood still dealing with the aftermath of a now-defunct oil refinery. This is a community that already knows precisely what industrial infrastructure does to a place long after the ribbon gets cut, now being asked to take a fresh set of promises about jobs and investment on faith, from a different industry making a very familiar pitch.

The context matters: data centres draw enormous amounts of electricity and water, and the people fighting these proposals generally aren't objecting to AI as an idea — they're objecting to the specific, physical thing turning up next door. The strain on the local grid, the water use, the construction traffic, and in cases like Philadelphia's, the fact that it's landing in places that have already paid an industrial price once and would rather not pay it twice. TechCrunch frames this as part of a wider outcry that's now spreading well beyond the usual data centre hotspots in the Sun Belt and Virginia.

What's actually happened here is that public opposition has stopped being a scattering of local news stories and become a documented national majority — and it's showing up hardest in exactly the neighbourhoods with the least patience left for round two. The Verge notes politicians are already responding to that same poll, which tells you this has moved well past a planning-committee nuisance and into something with real electoral weight behind it.

Right, this one's a bit more practical, and honestly a decent palate cleanser after all that. Meta has launched what it's calling an MCP server for WhatsApp Business. MCP stands for Model Context Protocol — think of it as the wiring that lets an AI coding assistant plug directly into a company's systems and actually do things inside them, rather than just chat with you about how you might do them yourself.

According to TechCrunch, this means developers can now point tools like Anthropic's Claude, the coding assistant Cursor, OpenAI's Codex, or ChatGPT itself at WhatsApp Business, and have those agents handle the setup directly: configuring the message templates a business needs for outbound texts, running the tests, and troubleshooting when something inevitably breaks. If you've ever tried to get a WhatsApp Business integration properly working, you'll know this is normally a genuinely tedious slog — template approvals, formatting quirks, testing flows nobody enjoys doing twice.

Some context on why this matters: WhatsApp Business is enormous, the default customer service and marketing channel for small and medium businesses across huge markets like India, Brazil, and Indonesia, and getting it configured correctly has always been a real barrier for developers without a dedicated integration team on staff. What's actually notable isn't the time-saving itself — plenty of tools promise that. It's that Meta built this specifically for AI agents to operate, not for a person clicking through a support dashboard. That's Meta making a concrete bet that a meaningful share of its developer traffic now arrives as an agent rather than a human, and wiring its infrastructure accordingly.

Which, once you notice it, starts showing up everywhere you look: steadily less software built for people to use, and more of it built for agents to operate on people's behalf.

So there we are — a chip boss with the President on speed dial insisting nobody needs supervising, a clear majority of the country that would rather not live next to the result, and Meta quietly building the plumbing on the assumption its next user won't be a person at all. Something in there for absolutely everyone's anxiety today.

And that's the news. A synthetic voice, a real source list, and a firm policy of not overstaying its welcome. See you tomorrow.