INAPP Blog

Facebook’s paying millions to creators who specialize in pissing people off. That’s not an exaggeration or a conspiracy talking point—it’s a line item in Meta’s regular disclosures since 2019. Last year alone, they cut checks to about two million accounts explicitly built around divisive, outrage-driven content. The math checks out: more engagement from the furious, more ads served, more revenue recycled into more of the same. The real kicker? We’re not talking about fringe accounts hiding in the dark corners of the platform. These are official “partner-publishers,” boasting verified checkmarks and revenue-sharing deals that let them turn hate clicks into hard currency. Meta’s own data shows the pattern: wherever they open their wallet, toxicity spikes and discourse curdles. You can trace the funding streams to the same playbook once reserved for conspiracy peddlers—only now it’s spreadsheets, not shadowy forums. I don’t know whether this is deliberate manipulation or just th...

I've been seeing some ridiculous stuff on LinkedIn lately, but this takes the cake: an AI tool that can generate cringeworthy posts in 8 different styles. You know, the kind of posts that make you roll your eyes so hard they get stuck that way. The tool's creators claim it can transform any topic into "peak LinkedIn cringe" - and honestly, they're not wrong. The styles range from "You Go Girl!" empowerment culture to "The Hot Take" opinion pieces, and even "Proud to Announce" shameless promotions. There's also "The Vulnerability Post", which I'm sure will be a thrill-ride of emotional manipulation. I mean, who doesn't love a good "I'm-so-vulnerable-but-also-so-inspired" post? The AI engine analyzes current trends to generate these masterpieces, and I have to admit, I'm a little impressed - not by the quality of the posts, but by the sheer audacity of creating a tool that can produce such clic...

Cider pulled pork with apple & sesame slaw Prep: 15 mins | Cook: 4 hrs 30 mins | Total: 4 hrs 45 mins | Serves: 4 There's something special about a warm, comforting dish like cider pulled pork with apple & sesame slaw that just can't be replicated by a restaurant or store-bought version. You know, the kind of meal that's been slow-cooked to perfection, with tender pork that just falls apart and a sweet, tangy slaw that adds a nice crunch. When you make cider pulled pork with apple & sesame slaw at home, you get to control the quality of the ingredients, like using a nice big pork shoulder, some caramelized onions, and a hint of smoked paprika. Plus, it's surprisingly easy on the wallet - a kilo of pork shoulder and some basic spices like caraway seeds and garlic granules are all you need to get started. As the pork cooks away in the cider, the aroma fills your k...

I'm still trying to wrap my head around the alleged misuse of Flock cameras by a New Bedford police officer to track her ex-partner. It's not just the fact that this happened, but how quickly the police department responded once the story started to unravel - they temporarily suspended their use of Flock cameras just a day after being questioned about it. The whole thing raises some serious concerns about surveillance technology and accountability. The details of the case are pretty disturbing. Officer Emily Pacheco is facing an internal affairs investigation, and she was served with an abuse prevention order just this Monday. What really caught my attention, though, was the police department's swift response to the inquiry. It makes you wonder if they were already aware of the potential for misuse, or if they're just trying to contain the fallout. As I dug deeper into the story, I started to think about the broader implications of surveillance technology and h...

I've always thought of data compression as a clever trick, a way to squeeze more information into a smaller space. But it turns out that's not entirely accurate. A new approach to compression has revealed that the fundamental principle behind it is actually prediction, not just squeezing. This changes everything, because if you can accurately predict what's coming next in a stream of data, you can represent it using much less information. What's surprising is how this realization can upend our assumptions about efficient data storage and transmission. For years, we've been focused on finding better ways to compress data, without really questioning the underlying mechanics. But if prediction is the key, that means we need to rethink our entire approach. It's not just about finding more efficient algorithms, but about understanding the patterns and structures that underlie the data itself. This has significant implications, because if we can get predictio...

Palm Springs Date Shake Total: 15 minutes | Serves: 2 Servings The first time I ordered a Palm Springs Date Shake, I braced myself for something aggressively sweet—a sticky, syrupy smoothie, maybe. What arrived was completely different: thick, cold, and quietly rich, with a faint nutty edge I couldn’t place. The counter guy finally pointed to the tahini. Tahini in a date shake? That tiny revelation changed how I thought about this desert classic. This version recreates that moment. Creamy vanilla ice cream gets whirled with six soft Medjool dates, a heaping spoonful of tahini, and a good pinch of salt to keep everything balanced. The cinnamon is optional but lovely, adding a subtle warmth that makes the shake feel like a Palm Springs morning. I’ve made this Palm Springs Date Shake countless times since, and every sip still takes me back to that first delicious surprise. It’s not complica...

I’ve never been happy with the idea of locking up model reasoning in encrypted blocks. If we’re going to treat an AI’s internal thoughts like state secrets, we’d better be damn sure they stay secret. The latest research suggests we might not be. Researchers showed how to smuggle sensitive data straight out of proprietary LLM APIs by hiding it in the encrypted reasoning blocks that models return. It’s not some elaborate side-channel attack—just a clever abuse of how providers handle their own encryption. They demonstrated this across OpenAI, Anthropic, and Google’s latest models, which should make anyone running a commercial API pause. If you’re trusting these blocks to keep secrets while they shuttle between user and server, you’re trusting the wrong thing. The kicker? This isn’t just theory. They tested it with a sanitized GitHub repo search—no API keys, but plenty of tokens that don’t belong in public code. The results were messy, but the method worked. And if it works once,...