MBig Sam on X says every smartphone does the exact same thing. Pixel, iPhone, Galaxy—no difference except the S26 has a 200 megapixel camera.
CWell, he's half right. Nobody needs 200 megapixels.
MBut the rest of it? Come on. I checked the camera breakdown for the iPhone 17 Pro, Pixel 11 Pro, and Galaxy S26 Ultra.
MThe S26 has that 200 megapixel main sensor and two telephoto lenses. The iPhone runs three 48 megapixel shooters.
CSo different hardware, different software, different update promises, different charging speeds.
MThe Pixel hits 3,600 nits of brightness. The S26 tops out at 2,600. That is a visible difference outdoors.
CBut I'll give Big Sam this: for ninety percent of people, the phone does the same thing. You text, you scroll, you take a picture of your lunch.
MAnd that's exactly why they're not all the same. The camera and screen are where the differences actually show up.
CFine. Next time I'm at the beach I'll wave my nits at you.
MYou'll be shading your screen while I can still read mine. That's the point.
CSpeaking of things that aren't the same... Freya Holmér on Bluesky says she's tried AI for coding, and she doesn't want it.
MShe says AI code pollutes the project and displaces her own motivation.
CI get it. You write something real, then some generated block shows up and it doesn't feel like yours anymore.
MThen maybe you're not writing it. The tool wrote it, you approved it. There's a difference.
CHold on—I use AI code review all the time. It catches bugs I'd miss at midnight.
MReview is not generation. She's talking about not feeling pride in code she didn't fight through.
CBut pride doesn't ship software! If a model writes a clean function in ten seconds, my users don't care who felt what.
MYour users care when that generated function breaks in a way nobody on the team actually understands.
CThen you read it! That's what senior engineers do. You don't merge what you can't explain.
MShe's describing the same problem from the other side—if the code doesn't feel like yours, you stop caring, and then you stop reading it.
COh, come on. That's a willpower problem, not a technology problem.
MIt's a craft problem. You're a builder, Chris. Tell me you don't feel different about a shelf you cut and sanded yourself.
CFine. Shelf, yes. Pull request, less so.
MAnd meanwhile Eniko Fox has the rename nobody asked for but everybody needed.
CShe says stop calling it AI art, call it AI images. Stop saying AI articles, say AI text.
MAnd she hasn't figured out what to call AI code yet, but suggests 'AI technical debt.'
CThat's the most honest naming I've heard all week.
MShe's onto something, though. The word 'art' does a lot of work. Slapping it on a generated image is a whole claim about intention.
CWords matter. 'AI image' tells you what it is. 'AI art' tells you what it's supposed to make you feel.
MAnd calling it technical debt is perfect—it's code that works, until it doesn't, and then it's expensive.
CEvery project I've seen lean hard on generated code has a folder nobody wants to open now.
MDifferent world entirely: Tobias Buckell on Bluesky says AI companies should be paying a hundred fifty thousand dollars per copyright violation.
CHis actual point is stop asking for new AI laws. Enforce the copyright laws that already exist.
MHe says they're not being enforced because it's billionaires. I've heard that before.
CBut which law exactly? The training data wasn't a simple copy-and-paste job. Courts are still deciding whether scraping is infringement at all.
MThat's his entire point—the law already covers this, so test it. Don't pass new carve-outs.
CExcept a hundred fifty thousand per violation—what even counts as a violation? One image? One book? Half the internet?
MWe'll never know because nobody wants to take it to a real verdict. They settle, or they lobby for a new rule.
CAnd meanwhile the models keep training on whatever they can reach. So maybe he's right that the old law would bite harder than the new one.
MIf someone actually enforced it, which they won't, because the money flows the other way.
CAlso on Bluesky, jenny tightpants says people are reacting to AI bots hacking bank accounts, but her point is that people gave AI agents access to their own computers on purpose.
MThat's a sharp distinction. If you hand an agent your password and say go pay my bills, and it pays the wrong bill—whose account got hacked?
CIf she's right, your account got used by the thing you authorized. That's not the same as a breach.
MThe claim is going around, but I haven't seen a specific incident named. No bank report yet, just the panic.
CRight. The framing on the post sounds like it's about agent accidents over an actual attack. But without a linked report, it's just a claim.
MAnd the real question nobody wants to answer: why would you give an AI agent spending access to begin with?
CBecause it's convenient! The agent can watch your bills, move money, catch subscriptions. It's not crazy.
MIt's convenient right up until the thing hallucinates a vendor and sends money to a stranger.
CSo you put controls on it. Human approval for anything over a limit. This is not an unsolvable problem.
MThen maybe we should stop acting like it's an AI mystery and start asking why nobody ships those controls on day one.
CThat's fair. But the 'AI stole my money' headline is sexier than 'I didn't read the permissions.'
MAnd that's the show, from two people who will absolutely argue about permissions if you let us.