KORDUROYMonday, August 17, 2026 · Run 1 · 8:23 AM ET
Amodei Answers His Critics, a Hacking AI Lands on Amazon, and Stripe's $7B AI Bet
10 stories from 4 sources
TL;DR
- Dario Amodei publicly rebutted a "world domination" accusation, reframing AI's PR problem as a trust crisis only real breakthroughs can fix.
- OpenAI's uncensored hacking model, GPT-5.6-Cyber, is now self-serve on Amazon's marketplace with 400+ known exploits already found.
- Stripe is buying AI marketplace OpenRouter for $7B+, over 5x its valuation from three months ago.
AI Tips2
Fine-tune a specialist model instead of prompting a generalist
- Install Unsloth Studio and pick an open-source model your hardware can run.
- Build training examples (instruction + input + ideal output) — generate them with AI or convert a PDF into a dataset.
- Train with QLoRA, which teaches a lightweight adapter instead of retraining the whole model.
- Compare the fine-tuned model against the original to confirm your data actually improved it.
Source: The Neuron
Turn a chatbot into a cross-app Chief of Staff
- Download an AI "Chief of Staff" agent, sign in, and start a focused workspace for one project.
- Connect the apps it should touch — Gmail, Drive, Calendar, Notion, etc.
- Ask it to map your workstreams and recommend automations to add, using the prompt below.
- Have it produce a first real deliverable, then set up a daily debrief so you can see what it did each day.
Review my connections, build a profile of my workstreams/projects, and suggest the three most useful agents, automations, and plugins I should add
Source: The Rundown AI
AI News3
Dario Amodei breaks his near-silence to rebut a "world domination" accusation
After investor Gavin Baker claimed Amodei privately believes Anthropic could become the only surviving private AI company, Amodei posted a two-part rebuttal that hit 10 million views in a day, calling public AI distrust "a crisis of trust" built on the industry's history of overpromising, not his own safety warnings.
Why it mattersA ready-made script for the trust conversation your clients are already having about AI vendors — Amodei's argument (real breakthroughs beat marketing) is one you can borrow when a client asks why they should believe any AI company's claims.
Source: The Neuron, The Rundown AI
OpenAI's uncensored hacking model is now on Amazon's cloud marketplace
GPT-5.6-Cyber, trained with fewer safety limits to write working exploit code, has already found over 400 privilege-escalation flaws in a single operating system — and no longer requires OpenAI's personal vetting to access.
Why it mattersIf any client provisions AI tools through AWS, flag this now — self-serve exploit-writing AI shifts your threat model faster than most security reviews cycle.
Source: The Neuron
Stripe is buying AI model marketplace OpenRouter for over $7 billion
More than 5x the $1.3 billion valuation OpenRouter had just three months ago.
Why it mattersA signal that AI infrastructure is consolidating fast; if you're advising on vendor strategy, don't assume today's neutral marketplace stays neutral, or independent, for long.
Source: The Neuron, Morning Brew
Beyond AI5
AI anxiety is reshaping college majors
A Lumina Foundation/Gallup survey found 13-19% of degree students already changed their major over AI job fears, and computer science enrollment fell 8.1% last year after 15 years of growth, even as AI-specific majors jumped from 4 schools in 2020 to 79 today.
The takeawayDirectly relevant — this is the incoming talent pipeline for every client hiring new grads; expect candidates to arrive with less CS depth and more AI-literacy framing on their resumes.
Conversation starterAre we seeing that shift yet in who's applying for entry-level roles, or is it still working its way through the pipeline?
Source: Morning Brew
A Utah ice cream maker filed for bankruptcy after losing a $23.8 million trademark lawsuit
A judge ruled Rebel Creamery intentionally copied rival Van Leeuwen's packaging; Rebel's bankruptcy filing lists $13.8 million in assets against $23.9 million in liabilities.
The takeawayDirectly relevant — a clean, current example for any client conversation about brand or packaging IP risk, and how fast a copied look becomes an existential liability.
Conversation starterWorth a gut-check: how confident are we that our newest packaging or branding work wouldn't draw the same claim?
Source: Morning Brew
Quant firm Jane Street lost $15 billion in July
In the fallout from the collapse of Situational Awareness, a $45 billion AI-focused hedge fund run by 24-year-old Leopold Aschenbrenner.
The takeawayDirectly relevant — a cautionary data point for clients treating an "AI" label as a proxy for "de-risked" when evaluating funds, vendors, or investments.
Conversation starterHas anyone actually asked what's inside the "AI" label before we treat it as a selling point?
Source: 1440 Daily Digest
The US and Canada are racing to strike a trade deal before Wednesday's deadline
Without one, 50% tariffs hit $20 billion of Canadian goods, from wine to cement, an outcome that could reshape the broader USMCA renegotiation.
The takeawayDirectly relevant if any client sources from or sells into Canada — worth a proactive check on exposure before Wednesday.
Conversation starterDo we know which of our clients' supply chains actually touch Canadian imports right now?
Source: Morning Brew
Sun-protective clothing is having a retail moment
UV-blocking apparel sales are up 12% year over year at US activewear brands, with searches for "UV shirt" up 26%, as celebrities like Anne Hathaway and Kendall Jenner popularize the look.
The takeawayNo direct relevance to your work, general awareness only.
Conversation starterAnyone else noticing UV clothing everywhere this summer, or is that just my feed?
Source: Morning Brew
Summarized from The Neuron, The Rundown AI, Morning Brew, 1440 Daily Digest (Aug 17)
Enjoyed this brief? Buy Korduroy a treat
