🕶️ Meta and YouTube lose the addiction case

Anduril is sprinting toward a $60B valuation, but the guts look less Top Gun, more DMV. WIRED’s investigation details setbacks from burns to busted timelines—growth theater outrunning manufacturing.

The $30.5B defense upstart promised faster, cheaper missiles; instead, curtains hid crates, “Roberto” lagged, and inert-motor deadlines slipped. Safety caught up late while machines oozed and managers churned. See factory stumbles exposed.

VCs keep the IV open, and Ohio dangles big incentives. Ambition’s intact; throughput isn’t. Call it valuation-first industrialism—great headlines, rough schedules. Ambition meets reality.

Founder takeaway: Hardware scale = process, people, permits, SOPs—measured in quarters, not hype cycles. If you sell “faster, cheaper, better,” your Gemba must back the deck, or the market will.

Companies are quietly outsourcing life-changing decisions, like who gets hired, who gets a loan, or who gets medical care, to algorithms nobody fully understands. These “black box” AI systems are trained on mountains of past data, but the rules they follow are hidden even from their creators. That means they don’t just replicate human bias; they can magnify it, at scale, and without accountability.

In this video, we break down how these systems work, the infamous cases where they went wrong, and the growing regulations trying to rein them in. From Amazon’s failed AI recruiter to hospital algorithms that disadvantaged Black patients, the risks are real, and they’re already here. The question is: can we make AI fair, or are we just building a smarter way to discriminate?

They all run their businesses on beehiiv. Newsletters, websites, digital products, and more. beehiiv is the only platform you need to take your content business to the next level.

🚨Limited time offer: Get 30% off your first 3 months on beehiiv. Just use code JOIN30 at checkout.

Start building for 30% off today.

The tobacco-playbook metaphor for social media just stopped being a metaphor. A Los Angeles jury found the platforms negligent for addictive design and failure to warn, awarding $6M to a single user and explicitly citing “malice, oppression or fraud.”

Separately, a New Mexico jury ordered $375 million penalties under state child exploitation laws. Crucially, both cases sidestep Section 230 by attacking product mechanics—recommendation loops, notifications, filters—not user content.

This reframes “engagement” as a potential defect, not a metric. Plaintiffs now have a playbook: show internal awareness of youth harm, show designs optimized for time-on-platform, and argue failure to warn.

For any founder touching minors, mental health, or dark patterns, this is the new baseline risk. Expect discovery on growth experiments, safety reviews, and age gating, and more juries told these cases could open litigation floodgates.