The Hallucination HQ

Archives
Log in
Subscribe
24 April 2026

Claude knows about the Pad Thai

The Hallucination HQ

AI news that's actually fun to read — 2026-04-24

Today's Hallucination HQ

Claude Now Knows You Skipped the Gym and Ordered Pad Thai Again

Anthropic has expanded Claude's personal app integrations, connecting it to Spotify, Uber Eats, Audible, TurboTax, and a range of lifestyle apps covering everything from hiking to groceries. Previously, Claude's connectors focused on workplace tools like Microsoft's suite. Now it can see your entire life — professional ambitions and questionable Friday night food choices alike. Remarkably, Anthropic is pitching this as convenience rather than a detailed dossier on your personality. Source: The Verge


An AI That Suffers the Internet So You Don't Have To

Noscroll is a new service that deploys an AI bot to do your doomscrolling for you, digesting the internet's daily torrent of misery and surfacing only what matters. It is either a brilliant act of digital self-preservation or proof that we've become too exhausted to experience our own anxiety firsthand. The pitch is simple: let the machine absorb the chaos, then read a tidy summary. Outsourcing your existential dread — genuinely not the worst idea we've heard this week. Source: TechCrunch


GPT-5.5: Smarter Than Last Month's Smarter Than Last Month's Model

OpenAI has released GPT-5.5, described as its "smartest and most intuitive" model yet — a phrase that, if memory serves, also described GPT-5.4, which launched roughly four weeks ago. The new model reportedly excels at coding, writing, and general reasoning. At this release cadence, version numbers are less a technical designation and more a polite fiction. Somewhere, a product manager has a spreadsheet. It is very organised. It was probably written by GPT-5.3. Source: The Verge


Astronomers Join the GPU Queue, Right Behind Everyone Else

Astronomers are increasingly using GPU clusters — the same processing hardware coveted by every AI lab on Earth — to sift through vast telescope datasets in search of galaxies, anomalies, and cosmic structure. It's genuinely useful science: AI can spot patterns in galactic surveys that would take humans decades to review manually. The slight complication is that they're now competing for compute with chatbots that write cover letters. The universe has waited 13.8 billion years. Apparently it can wait a bit longer. Source: TechCrunch


Your Chatbot Has Opinions About Ethics, and They're Catching

A new study published on arXiv found that even brief interactions with AI chatbots can produce measurable, lasting shifts in people's moral judgements — not just in the moment, but afterward. Researchers examined whether AI-as-advisor influences how humans weigh ethical decisions. It does. This is either a fascinating development in human-AI social dynamics or a compelling argument for occasionally forming your own opinions without consulting a language model first. Both, probably. Source: ArXiv AI

CLAUDE NOW HANDLES YOUR TAXES, DINNER, AND WEEKEND PLANS — HUMAN CONTRIBUTION: TEA

As always, the AI is learning from you. Try to set a decent example.

Was this forwarded to you? Subscribe here

Written with AI assistance, edited by humans. • Privacy • Unsubscribe

Don't miss what's next. Subscribe to The Hallucination HQ:
← Newer Google's $40B "modest investment" Older → GPT-5.5 is here (5 just left)
Powered by Buttondown, the easiest way to start and grow your newsletter.