Oct 3, 2025 6 min read Daily Rundown

🛎️ Citi Trains 175,000 Staff

Plus: Fox Guards Henhouse, LLMs as Labor Brokers

Good Morning, AI Enthusiasts!

When an institution trains its entire workforce, when the guardians of labor invite disruption to redesign their own walls, and when algorithms begin to act as brokers of human effort, we see not separate events but a single pattern.

BANK

Citi Turns Staff Into Prompt Engineers

📌 What’s happening: Citi just mandated AI prompt training for all 175,000 employees across 80 countries. The internal memo calls it “the beginning of a new way of working,” where staff learn to write better prompts for the bank’s in-house AI tools. Employees have already logged 6.5 million prompts this year, and Citi wants to move from “basic prompting” to precision workflows. Competitors are on the same path: JPMorgan uses AI to generate investment pitch decks in 30 seconds and Wells Fargo shipped thousands to Stanford’s AI bootcamps.

🧠 How this hits reality: This is AI hitting compliance-grade scale, not pilots or hackathons but full-stack workforce retraining. Banks are effectively treating prompt fluency as a baseline skill like Excel or Bloomberg terminals. That resets hiring filters, rewrites onboarding, and forces middle managers to measure output in “AI leverage” rather than raw hours. If Citi can cut pitch prep or compliance tasks from days to minutes, then labor arbitrage shifts from geography to prompt literacy.

🛎️ Key takeaway: When 175,000 bankers are told to “prompt or perish,” AI isn’t an add-on, it’s the new Excel.

TOGETHER WITH QUILLBOT

QuillBot’s AI Humanizer instantly transforms AI-generated text into natural, engaging language. Just paste your content and hit “Humanize” to refine tone, improve flow, reduce robotic phrasing, and preserve meaning—all in seconds.

Try now👉

JOB

UK Jobs Office Pays IBM to “Fix” Unemployment With AI

📌 What’s happening: The UK Department for Work and Pensions signed a deal worth up to £27M with IBM under its Nexus AI program. Big Blue is tasked with building prototypes, scanning horizons, and deploying machine learning into welfare systems. On paper it’s about efficiency; in reality it’s the office meant to help jobseekers hiring a technology widely feared to accelerate job loss.

🧠 How this hits reality: Handing welfare decisions to AI means the poorest citizens are judged by opaque models already accused of bias. The government frames it as “responsible, human-centered,” but efficiency targets shrink the space for empathy. The irony is brutal: the department for jobs is betting on the same tools blamed for eroding work. That’s less a safeguard than a fox guarding the henhouse.

🛎️ Key takeaway: Using AI to cure unemployment is like asking a weasel to watch the chickens—predictable, and never in your favor.

RESEARCH

When AI Plays Boss

📌 What’s happening: A McGill University study stress-tested wage-setting by AI, feeding 60,000 freelancer profiles into eight leading LLMs and generating 4 million pay recommendations. The method was clinical. Duplicate profiles were adjusted for a single attribute such as gender, age, or geography, then tested with prompt variations that instructed the model to consider, ignore, or explicitly weight those factors. The findings showed no gender bias, persistent age premiums, and geography gaps so large that a US location tag could instantly double the rate.

🧠 How this hits reality: The experiment shows that AI is already shaping labor markets, not just task outputs. If algorithms start setting pay by default, we’re essentially outsourcing wage policy to models trained on messy human history. And just like the dream of “scientific communism,” perfectly fair AI wages remain a utopia—data bias is stubborn, context is messy, and prompts can’t fix everything.

🛎️ Key takeaway: Teaching AI to pay workers fairly may be as hard as engineering communism—possible on paper, broken in practice.

QUICK HITS

The US National Park Service is testing AI-powered autonomous lawn mowers at six national parks, including the National Mall.
Axiom Math raised $64 million at a $300 million valuation to build an AI mathematician that creates and proves new math.
Queens-based dispensary TRENDS became the first in the U.S. to launch an AI-powered platform to guide cannabis wellness choices.
Women’s health startup Midi Health raised $50 million, betting on longevity and AI to bring menopause care mainstream.
AI-generated character Tilly Norwood landed a Wikipedia page that sparked heated debate among editors over whether she can be called an “actress.”

TRENDING

Daily AI Launches

Instruct has launched a "truly no-code" platform where you can build, edit, and run powerful AI agents using only natural language.
Caesr AI has launched as an AI agent that automates tasks on any device by clicking, typing, and navigating for you based on a simple prompt.
The Merge Agent Handler has launched to securely connect AI agents to enterprise tools via MCP.
Dreamlit AI has launched a tool that lets you create beautiful, automated email workflows by "vibe coding" them with natural language prompts.

Featured AI Tools

💬 CREAO makes future apps possible for all.
📚 Heardly is the Fast Way to read Best Book.
🪶 CopyOwl is the First AI Research Agent, deep research on any topic in one click.

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, boasting over 1 million readers from leading companies such as OpenAI, Google, Meta, and Microsoft. Our Newsletter Brands: