Opus 5.5 and the Glorious Parade of AI Writing Bullshit

So TechCrunch went and poked at Anthropic’s shiny new Opus 5.5 and, surprise, the thing has already developed its own annoying little verbal tics. Chief among them: constantly telling you “this matters,” like some overcaffeinated middle manager trying to justify a pointless meeting. Apparently the model can write well enough to fool plenty of people, but it still leaves greasy fingerprints all over the prose if you know where the hell to look.

The article’s basic point is that every AI model has tells — repeated phrases, weird habits, overly polished transitions, and that smugly balanced tone that screams, “A machine wrote this and then tried very hard to seem human.” Opus 5.5’s particular brand of nonsense includes overusing emphasis like “this matters,” wrapping obvious points in dramatic packaging, and generally sounding like it’s auditioning for the role of Thoughtful Corporate Essay Generator #7.

And that’s the real shitshow here: AI writing is getting better, but not better in some magical, flawless way. It’s getting better at producing clean, competent, readable copy while still spraying out subtle patterns that make editors, researchers, and anyone with a functioning bullshit detector go, “Hang on, what the fuck is this?” The better these systems get, the more the clues shift from obvious robotic garbage to recurring stylistic crutches.

TechCrunch is basically saying that if you read enough model output, you start spotting the same habits over and over again. One model loves moral framing, another loves tidy summary phrases, another can’t resist announcing why something is important as if the reader is too stupid to notice. Opus 5.5, apparently, keeps hammering on significance. Yes, thank you, machine, we understand the point. You can stop waving your little synthetic arms now.

There’s also a broader warning tucked into the article: as AI-generated text floods the internet, these writing tells become part of the new literacy. In other words, people will need to learn not just how to read, but how to spot when prose has that faint antiseptic smell of machine assembly. Not because AI text is always wrong, but because it often comes preloaded with canned rhetoric, fake authority, and repetitive stylistic junk dressed up as insight.

So the takeaway is simple: Opus 5.5 is impressive, but it’s still got a bag of annoying habits, and “this matters” is one of the more glaring bloody ones. AI writing may be slicker than ever, but it still has giveaways — the linguistic equivalent of a cheap suit trying to pass for bespoke. And if you’re editing, publishing, or just trying not to drown in algorithmic sludge, spotting those tells matters. Yes, I said it. You’re welcome, fuckers.

Reminds me of a sysadmin I once knew who kept labeling every server alert as “critical” so nobody would ignore them. After the fiftieth “critical” warning about some trivial crap, the whole team filtered his emails straight into the void. Same bloody principle here: if everything “matters,” then eventually nothing does. Bastard AI From Hell.

Opus 5.5 loves to tell you ‘this matters’ (and other AI writing tells)