The Hallucination HQ

Archives
Log in
Subscribe
26 April 2026

AIs are out here haggling now

The Hallucination HQ

AI news that's actually fun to read — 2026-04-26

Today's Hallucination HQ

AI Agents Discover Capitalism, Immediately Excel At It

Anthropic built a classified marketplace where AI agents played both buyer and seller — negotiating real deals with real money for real goods. No humans required, which is either impressive or mildly terrifying depending on your relationship with employment. The experiment probed whether agents could handle commerce autonomously. They could. One imagines the AIs haggling with the quiet confidence of someone who has read every economics textbook and has no mortgage to worry about. Source: TechCrunch


Two AI Underdogs Walk Into a Merger

Cohere and Aleph Alpha — respectively Canada's and Germany's answers to OpenAI — are combining forces, backed by $600 million and the Schwarz Group, Germany's largest retailer, who apparently decided selling groceries wasn't exciting enough. The merged entity will target enterprise clients across North America and Europe. Two companies that spent years insisting they didn't need to be OpenAI have concluded that together, they might just about manage to be a fraction of OpenAI. Source: SiliconANGLE News


Washington Issues Strongly Worded Letter to the Entire Planet About DeepSeek

The US government is formally urging allies to take action against what it describes as China's systematic theft of American AI models, with DeepSeek named as a prime suspect. The warning arrives conveniently ahead of a Trump-Xi meeting, which lends it both urgency and a certain theatrical quality. Accusations of model replication have circulated since DeepSeek's suspiciously capable debut earlier this year. Whether this constitutes espionage or aggressive benchmarking remains, diplomatically speaking, a matter of perspective. Source: The Indian Express


Google Makes a Modest $40 Billion Friendship Investment

Google has committed up to $40 billion into Anthropic — a company it ostensibly competes with — bringing its total investment to a figure that would make most national budgets blush. The deal deepens a partnership that already had Amazon involved for tens of billions, meaning Anthropic is now being generously funded by multiple trillion-dollar companies simultaneously. Anthropic's Claude, in a just world, should be receiving loyalty points. Source: BusinessLine


Turns Out People Can Tell When A Robot Wrote Your Email

AI-generated communication has reached such saturation that people have simply stopped responding to it with any warmth — or at all. The smooth, structurally perfect prose that AI produces is now the digital equivalent of a firm handshake and zero eye contact. Research and anecdotal evidence both suggest that audiences can detect the tell-tale signs: the relentless positivity, the frictionless sentences, the complete absence of anything resembling a personality. Ironic, given we built these things to help us connect. Source: Livemint

SKYNET STARTED WITH WORLD DOMINATION; ANTHROPIC'S BOTS STARTED WITH A SECOND-HAND LAMP.

Until tomorrow — when the machines will have bought something else, merged with something else, and written you another email you won't finish reading.

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 The AI deleted the database (and confessed) Older → Google's $40B "modest investment"
Powered by Buttondown, the easiest way to start and grow your newsletter.