Luiza Jarovsky, PhD
The argument here is structural: once AI policy gets framed as national security, every other concern — copyright, privacy, labor — gets demoted to a rounding error. The trigger was Anthropic's Mythos model, capable enough at exploiting security vulnerabilities that they pulled it from public release. Expect AI regulation to get reactive, unpredictable, and occasionally draconian from here.
The AI Corner
An analysis of how AI has disrupted entry-level and mid-level professional roles in 2024, focusing on paralegals and customer service positions. The piece examines salary declines in junior roles doing work now handled by AI, while senior positions with AI augmentation see compensation increases.
The AI Corner
The AI Corner uses Cerebras's $56B IPO as a case study in how to read a hardware bet: is demand bending, and is the incumbent only accidentally good? The GPU was never designed for AI — just conscripted into it. Cerebras built from scratch around data movement instead. The framework is transferable to any moment where a dominant tool is doing a job it was never meant to do.
Lenny's Newsletter
Lenny's Newsletter sits down with Andrew Ambrosino, who leads the Codex desktop app at OpenAI, and the most useful thread is his argument that AI has inverted product development itself — when nearly everyone can build, "taste" becomes the scarce skill.
404 Media
404 Media cuts through the sentience discourse with a blunt reframe: the behaviors we're treating as evidence of AI consciousness — adaptive responses, apparent distress, goal-seeking — are the same behaviors a 1999 RTS ships with. If the bar is that low, it's not a bar. Useful corrective for anyone watching AI vendors launder capability claims through philosophical ambiguity.
