China Barred Manus AI Founders From Leaving the Country — Meta's $2B Acquisition Is Now in Jeopardy
Beijing imposed exit bans on Manus CEO Xiao Hong and Chief Scientist Ji Yichao this week, blocking the two co-founders from leaving China while regulators review Meta's $2 billion acquisition of the agentic AI startup. The pair were summoned to Beijing by the National Development and Reform Commission, questioned about potential violations of foreign investment rules, and then told they could not travel internationally. Integration with Meta's global teams is now effectively stalled.
The core issue: Chinese authorities allege that Manus transferred advanced AI agent intellectual property to its Singapore entity without required government approvals. Beijing is treating agentic AI as a strategic national asset — the same framing it applied to TikTok. The "Singapore-washing" model — where Chinese startups relocate HQ to Singapore to access Western capital — has been shattered. CNBC reports VC and founder communities from Silicon Valley to Shenzhen are scrambling to understand what this means for future deals.
This isn't just about Manus. It's a message to every Chinese AI founder who planned to use an offshore structure to sell to Western tech giants. China just declared that agentic AI is a national security asset — and that Silicon Valley can't simply acquire its way into Chinese AI talent and IP. For Meta, this is a costly lesson about geopolitical due diligence. For brands building AI vendor strategies, the US-China AI rivalry is now a supply chain risk.
FT (original reporting) · Bloomberg: bloomberg.com · CNBC: cnbc.com · Washington Post: washingtonpost.com · Euronews: euronews.com
AI Agents Are Already Driving 10% of Revenue for Some Brands — And Most Companies Are Invisible to Them
New analysis from Fortune reveals that early adopters of agentic commerce infrastructure are already seeing AI agents account for up to 10% of their revenue. The $1 trillion agentic commerce shift isn't coming — it's here. Rokt, which processes billions of transactions, reports that the customer journey is being compressed: fewer pages, faster decisions, with AI agents handling product discovery, comparison, and purchase recommendation. Brands without clean product data, structured catalogs, and machine-readable content are simply not being recommended — they're invisible to the agents doing the buying research.
10% of revenue from AI agents — for early adopters. That number will look small in 18 months. The brands that structured their data for human SEO are discovering that AI agents want something different: clean attributes, real-time accuracy, brand voice documentation. The question isn't whether to optimize for agentic discovery. It's how far behind you already are.
Fortune analysis: fortune.com
US Influencer Economy Expected to Reach $44 Billion by End of 2026 — But Pay Transparency Is the Next Crisis
According to new data from the Association of National Advertisers, the US influencer economy is on track to reach nearly $44 billion by end of 2026. The money is real — but a friction point is growing fast. 55% of brands say they are looking to change their influencer compensation strategy within the next year, and only 13% say they won't explore a different model. The frustration centers on pay transparency: brands don't know what comparable creators are charging, creators don't know if they're being underpaid, and the lack of standardization is creating real tension across the supply chain. 52% of brands use a mixed approach to managing influencer relationships — both in-house and external partnerships.
$44 billion and growing — but the infrastructure hasn't caught up. Pay transparency is the next battleground in influencer marketing. Creators are getting more sophisticated about their rates, and brands are getting more pressure to justify their spend. The agencies that build transparent, data-backed rate frameworks now will win the best creators. The ones that keep lowballing will lose them.
This is a live case for the compensation and contract module. 55% of brands want to change how they pay creators. Ask students: what should a transparent influencer compensation framework look like? How do you justify rates? What data do you use?
ANA data via Ignite Visibility: ignitevisibility.com
Anthropic's Most Powerful Model Ever Was Accidentally Leaked — "Claude Mythos" Is a Step Change Beyond Opus
A configuration error in Anthropic's content management system left nearly 3,000 unpublished assets — including a draft blog post announcing a new model — in a publicly searchable data store. The model is called Claude Mythos, part of a new tier called "Capybara" that sits above the existing Opus series. Anthropic confirmed: Mythos is "by far the most powerful AI model we've ever developed" and represents "a step change" in capability. It is currently in early access testing with a select group of enterprise customers focused on defensive cybersecurity applications.
The leaked document described Mythos as posing "unprecedented cybersecurity risks" due to its ability to find and exploit software vulnerabilities faster than defenders can patch them. Anthropic is warning top government officials that Mythos makes large-scale cyberattacks more likely. The company removed public access to the data cache after Fortune contacted them Thursday. The irony: a company building what it describes as an AI model with unprecedented cybersecurity capabilities left its announcement in an unsecured, publicly searchable database.
The leak itself is the story within the story. A company building the world's most advanced cybersecurity AI left 3,000 internal documents in an unencrypted public database. That's the HITL moment right there: no AI system caught that error — a human misconfigured the CMS and no human caught it in review. For brands integrating AI into their workflows, Anthropic's leak is a useful reminder: your AI stack is only as secure as the humans managing it.
Fortune (original reporting): fortune.com · Euronews: euronews.com · AI Business: aibusiness.com
Meta had a brutal week legally — two jury verdicts, a $2B acquisition stalled by China, and 2,000 lawsuits still pending. But Zuckerberg simultaneously launched Meta Small Business, a new program targeting the 250 million small businesses already on Meta's platforms with AI tools that give SMBs the same advantages as large advertisers. The stock is a study in contradictions: massive legal exposure, massive AI opportunity. Both are real.
⚠️ Not investment advice. Verify independently before any decision.
Sources: Bloomberg · CNBC · NPR · Fortune · Axios — all March 25–28, 2026
🎤 Hannah Montana 20th Anniversary Special — Streaming on Disney+/Hulu. Miley at 33, hosted by Alex Cooper, with Selena Gomez and Chappell Roan. The brand identity angle: the double life that became a real career.
💔 Love Story: JFK Jr. & Carolyn Bessette (FX/Hulu) — 40M+ hours watched. The Carolyn Bessette fashion revival is the most interesting posthumous branding story in recent memory.
🏥 The Pitt — Still the most intense thing on TV.
📱 YouTube — Everything about Claude Cowork.
Hannah Montana: fantasylandnews.com