A researcher in AI security sits down Monday morning. Over the weekend, two things happened: a new paper on jailbreak attacks landed on arXiv, and an AI lab pushed a model update that changes the relevant attack surface. Neither made it into her inbox.
The arXiv paper was posted Friday afternoon. Email notifications for that day's arXiv submissions had already gone out, and the next batch won't arrive until Monday morning. The model announcement appeared in a lab blog post, but she subscribes to a newsletter that covers AI news on Tuesdays. She'll find out when a labmate mentions it, when she goes looking herself, or at submission time when a reviewer asks why she didn't address it. By then she may have written two sections that need reframing.
This is the default condition for researchers who work across multiple fast-moving fields.
For most of the history of academic publishing, staying current was domain-specific. You read the journals in your area, attended the right conferences, subscribed to a handful of mailing lists. The field moved at a pace those tools could track. That model started breaking down when AI development shifted from primarily academic to primarily industry-driven. Labs now publish blog posts and technical reports more often than formal papers, and those announcements matter as much as the papers do.
The relevant information for a researcher in adversarial machine learning now spans arXiv preprints, AI lab announcements (no standard venue, no consistent schedule), security advisories when deployed models get attacked, and the startup ecosystem that determines which models end up in production systems. These sit in completely different channels, and nothing was built to cover all of them.
Existing tools don't fill the gap. RSS readers work if you maintain them; most people abandon them within a few months because the signal-to-noise ratio degrades without a curation layer. Subscribing to ten newsletters means you're doing the aggregation yourself in your inbox every morning. Academic Slack channels surface community discussion but not systematic coverage. Social media worked for tech discourse until the main venues for it fragmented.
What's missing is daily curation built around how interdisciplinary research actually works: industry-adjacent, crossing domains that don't share a publication venue, moving faster than a journal cycle. Not a general tech newsletter. Not a pure paper feed. Something that tracks the combination of fields you care about and synthesizes it into something readable before your first meeting.
The Purplelink Daily Digest is my attempt to build that for myself. It's open to anyone who reads in a similar way.