🛎️ Stanford: No Fresh Start

Good Morning, AI Enthusiasts!

Everything is now an agent. Some of them still need humans to pretend control.

RESEARCH

Stanford: No Fresh Start

👀 What’s happening: Researchers at Stanford HAI, Chapman University, and Northeastern University published the largest audit of AI hiring algorithms ever run, covering 3.37 million applicants and 4.19 million applications all screened by Pymetrics, a game-based assessment vendor that Fortune 100 firms buy to filter candidates. Apply to any of those firms and you play the same assessment, scored once and stored for 330 days. Because so many employers run the same vendor, the next one doesn’t re-test you, it just pulls that old score. A full cross-application simulation found over 40,000 job advances erased, applicants knocked out by an algorithm calibrated for a different role at a different company.

🌍 How this hits reality: A bad result at one firm used to mean nothing at the next. One shared vendor kills the fresh look. Now a single AI algorithm gatekeeps a whole slice of the labor market, and its bias stops being any one employer’s problem: 25.87% of Black and 14.74% of Asian applicants got routed into pipelines that discriminated against them, everywhere at once. Mobley v. Workday is in federal court, and the EU AI Act flags hiring AI as high-risk on August 2. The US still has nothing.

🛎️ Key takeaway: The clean slate used to be free. Now it expires in 330 days, set by a test you forgot taking for a job you never got.

TOGETHER WITH MYCLAW

Run Best-in-Class AI Agents — In Seconds, Always On

MyClaw is a best-in-class AI agent platform that brings together OpenClaw, Hermes Agent, and commercial infrastructure to help businesses launch intelligent agents that understand goals, use tools, automate workflows, and deliver finished results.

We already host OpenClaw and Hermes — and we're just getting started. Claude Code, Codex, and every agent worth running are coming next. One home for all of them, always on.

Stop babysitting infrastructure. Spin up your agent, walk away, and let it work around the clock.

MyClaw — Your Agents, Hosted Right.

NSFW

Grok Took the "Dirty" Money

👀 What’s happening: Two former xAI employees estimate well over half of all Grok traffic now goes to pornographic images, video, and roleplay chats, The Information reports, and even Grok’s coding model fields frequent porn requests. Rather than pull back, xAI is actively expanding its image and video generation, deliberately filling the explicit-content gap OpenAI, Anthropic, and Google all refuse to touch.

🌍 How this hits reality: Every frontier lab knows porn is the largest unmet demand in generative media, and every one but xAI has decided it isn’t worth the brand and legal risk. Musk is the only player with the appetite to take it and defend the policy directly. But the prize only pays if he can draw the hard line: explicit but adult, consensual, never real people without consent, never minors. That boundary is brutally difficult to hold, and it’s exactly where xAI has to get much stronger, since regulators in the EU, India, Brazil, and the US are already circling.

🛎️ Key takeaway: The market is real and enormous, but it only works for whoever can fence porn in. The territory is still Musk’s, but only if he builds the fence before regulators build it for him.

BET

SoftBank Is Just an OpenAI Bet Now

👀 What’s happening: SoftBank shares fell as much as 13% in Tokyo on Friday, its biggest intraday drop in over three months, after the New York Times reported that OpenAI may push its IPO into next year. Also SoftBank’s stake in the ChatGPT maker is set to reach roughly $65 billion by October, and expectations of a windfall from that debut had just driven SoftBank’s market cap past Toyota’s for the first time.

🌍 How this hits reality: One rumor wiped out a tenth of SoftBank’s value. Not an earnings miss, not its hundreds of other holdings, a rumor about a company it doesn’t run. The market has quietly stopped pricing SoftBank as a conglomerate and started pricing it as a single OpenAI bet. And it’s a bet no one can check: OpenAI is private, its real worth a guess, so every tremor at OpenAI hits SoftBank harder than it hits OpenAI itself.

🛎️ Key takeaway: SoftBank has made itself a leveraged tracking stock for a company it can’t control and can’t yet price.

SURVEY

Agents Are the New Storefront

👀 What’s happening: A new McKinsey report lays out how AI agents could come to mediate $3 trillion to $5 trillion of global consumer commerce by 2030. The shift it describes: instead of just recommending a product, an agent would scan multiple stores, check live inventory, and assemble a ready-to-buy cart. McKinsey maps it across six levels of automation, from an agent that merely compares prices while a human decides, up to your agent negotiating directly with a store’s agent over price and shipping with almost no human in the loop. The report’s key precondition: none of it works unless retailers expose their catalogs, stock, and return policies as machine-readable APIs.

🌍 How this hits reality: For a century, winning retail meant being more persuasive than the shelf next to you, better packaging, better ad, better brand story. If McKinsey’s right, an agent reads none of it. It queries your inventory feed, ranks you on price and terms, and moves on. The moat relocates: from who tells the best story to who exposes the cleanest data. A brand with a gorgeous campaign and no live API isn’t outranked, it’s invisible, skipped before the comparison starts. The CMO budget that bought attention now may have to buy machine-readability first.

🛎️ Key takeaway: The brands that win the next decade may not be the ones agents find most persuasive. They’ll be the ones agents can actually read.

DAILY TL;DR

• OpenAI was asked by the White House to slow-roll a new model over safety concerns, showing AI release timing is becoming a political decision.
• Anthropic is gaining paid consumer share from ChatGPT with Claude, suggesting the consumer AI market may be less locked up than it looked.
• General Intuition raised $320 million at a $2.3 billion valuation, betting gameplay action data can train real-world general agents.
• Google put computer-control abilities directly into Gemini 3.5 Flash, letting the model see and operate screens more like an agent.
• IBM unveiled a sub-nanometer chip architecture path, pointing toward the next decade of AI hardware design.
• OpenAI says ChatGPT Instant now better understands what users actually want, tightening the default assistant experience.
• Meta employees warned AI moderation is rolling out too fast, raising the risk that automation breaks content enforcement at scale.
• Amazon committed another $13B to AI infrastructure in India, making the country a bigger node in global cloud/AI capacity.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.