Hey everyone,

This is the first issue where I’m thinking of Late Stage less as a single-project newsletter and more as a signal feed from the edge of AI, web3, security, and the strange digital worlds forming between them.

Late Stage Archetypes is still the center of gravity; the archive, the characters, the signal fragments, the dystopian identity layer.

But I’m also building something adjacent: Wraith, a hands-on AI security academy where people learn to attack and defend LLM agents through live CTF-style challenges.

The overlap is starting to feel obvious.

One project explores identity inside collapsing systems.
The other trains people to break the systems that are now being built around us.

Different layers. Same world.

The Late Stage Archetypes drop is now less than 100 days away.

The collection is built around a universe of corrupted portraits, reconstructed identities, and strange archetypes recovered from a failing archive. Humans, Undead, Aliens, Apes — each one treated less like a “trait stack” and more like a recovered signal.

I’ve also been continuing to build around the broader Late Stage world, including Signal Grid, the playable layer at:

play.latestage.app

Signal Grid is not “the utility” for the NFT collection, and I don’t want to over-promise anything there. I think of it more as an early playable artifact from the universe — a way to interact with the tone, mechanics, and mythos before the main collection opens.

The NFT drop remains focused on the art, the archive, the characters, and the collector network forming around it.

More lore updates, archetype reveals, and allowlist mechanics are coming soon.

I’ve also been building and launching Wraith:

wraith.sh

Wraith is a free, hands-on AI pentesting academy. The core experience is CTF-style challenges against live LLM agents, prompt injection, tool abuse, data exfiltration, agent manipulation, and other attack patterns that are becoming more relevant as AI systems start touching real workflows.

The certification is called:

WCAP — Wraith Certified AI Pentester

The normal exam price is listed at $299, but during launch I’m currently offering a free WCAP exam attempt to anyone who completes 5 Wraith challenges.

That offer is temporary, but for now it is meant to reward early users and help seed the first wave of certified Wraith players.

There are already a handful of WCAP holders.

That still feels early. Which is good. Early is where culture forms.

AI is moving very quickly from “chatbot” to “agent.”

That means AI systems are no longer just answering questions. They are increasingly being connected to tools, files, browsers, inboxes, APIs, payment systems, codebases, and business workflows.

That changes the security model.

A normal chatbot can say something wrong.
An agent can potentially do something wrong.

That is the surface area Wraith is focused on.

And it is not just a niche technical concern anymore. The current AI news cycle is full of the same signals: more powerful models, more agentic workflows, more government attention, more infrastructure spending, and more concern about what happens when these systems are deployed too quickly.

Recent reports say the White House has been discussing a proposed framework where developers of high-risk frontier AI models could notify the U.S. government before major releases. Reuters reported that agencies may review some advanced models up to 90 days before public launch under a voluntary structure.

Google is also pushing deeper into agentic AI. Recent coverage of Google I/O described new agent-style products built around Gemini, Search, Gmail, Docs, and long-running background tasks; basically the direction everyone has been expecting: assistants that don’t just answer, but act.

Meanwhile, Anthropic co-founder Jack Clark recently predicted that AI could contribute to a Nobel Prize-winning discovery within a year, while also warning that the same acceleration creates serious risks if society remains reactive instead of prepared.

That combination: capability, autonomy, access, and uncertainty is exactly why AI security is going to matter.

Not someday.

Now.

The more I build, the more these worlds seem connected.

AI agents are becoming workers.
Digital collectibles are becoming identity objects.
Games are becoming distribution layers.
Security is becoming culture.
Culture is becoming infrastructure.

That is the Late Stage lens.

Not just “AI news.”
Not just “NFT updates.”
Not just “cybersecurity.”

More like: what happens when all of these systems start blending together?

That is what I want this newsletter to track.

Some issues will include project updates.
Some will include AI/web3 observations.
Some will include Wraith challenge notes, collector updates, lore fragments, or weird signals from the edge.

Thanks for being early.

The archive is still reconstructing.

— Anthony / WizardX

AI can do a lot, but most AI newsletters leave you wondering what’s actually worth trying. 

That’s where The Shift comes in. They don’t just tell you “what’s new” in AI. 

Every edition answers one question: What can you do with AI today that saves you time, money, or effort?

You’ll see real-world examples, step-by-step mini-guides, and instantly usable prompts under 5 minutes a day.

Plus free access to 3,000+ AI tools and 1000+ mega prompts so you can apply what you learn right away.

No hype, no wasted time, and no “just in case” news. If it’s in The Shift, it’s because it can make your work better today.

And right now, 3 subscribers win a free 1-year Claude Pro subscription. One click to enter.

Subscribe Now - It's Free!