Newsletter

Maligned

Weekly AI briefing. The important stuff, without the hype.

Latest Issue March 16, 2026

Maligned #14 - Agents, Video, and Practical Control

New work on secure AI agents, deep video understanding, and precise control for generative models.

Real-world AI deployment challenges surface, while specialised models gain traction and regulatory clarity starts to emerge.

This week, we look beyond raw scale to the practical advancements making AI usable and reliable in real-world applications.

This week, we saw smarter deployment strategies for ML, alongside crucial insights into how we truly measure AI progress.

AI is entering the part of the cycle where the hard, boring work actually matters.

Enterprise AI teams are sleepwalking into platform dependencies they will regret.

Multimodal AI moved from cool demos to actual production use cases this week.

The AI safety conversation is split between people doing real work and people performing for cameras.

The race to build smaller, more efficient models is producing results that matter for real deployments.

The copyright lawsuits are piling up and the outcomes will define the next decade of AI.

The EU AI Act enforcement timelines just got real and most companies are not ready.

AI agents are everywhere in pitch decks but scarce in production, and the gap is telling.

Everyone is fine-tuning models and most of them are wasting their time.

Cloud providers are slashing inference prices and it is about to reshape who can afford to build with AI.