Hardening security across every live module.
We are in the early weaponization phase of frontier AI. The team is reanalyzing the security assumptions behind every module we run in production. Surface by surface, layer by layer.
As we have discovered in the past, MultiversX, myself, and perhaps several other team members have been targets of a highly-sophistication attack campaign, by Lazarus and other groups among them. Every attempt has been detected, contained, defended. What is unmistakable to me, is that the raw intelligence now available to attackers has materially elevated the sophistication, speed, and creativity of these groups.
The vital work done by the team has surfaced a class of issues that did not exist as viable threat vectors twelve months ago. Each one identified, patched, and folded back into a continuous hardening pipeline.
Security is and continues to be the highest priority over the coming weeks. And the shift here is structural: raw AI offensive capability is being normalized into the baseline threat model of every open source project on earth. Until that single fact is internalized by open source and blockchain teams, new attacks and vulnerabilities will be exploited every single day. Responsible teams should swiftly reorder their priorities around this reality. Now.
Alongside the vital work done by our engineering team, we very much appreciate the continuous surgical effort put in by the resourceful network of builders, validators, and community members, surfacing issues, proposing improvements, and sharpening the system as we prepare for what comes next.
All focus is on this 360° security review until the hardening phase concludes.
Then, the next chapter.
SUPERNOVA.
100 security issues. Identified. Neutralized.
AI in its current form is a nuclear force. Channeled right, it delivers pure, unadulterated, unmatched, productivity gains. Channeled wrong, it ships weapons, with attack vectors the world has never had to defend against.
The next chapter will be defined by this duality: breathtaking productivity unlocks, colliding with deep, surgical, systemic attacks unlike anything we've seen.
For any blockchain team, any open-source team, security is THE PRIORITY.
The MultiversX team has been operating around the clock. Nights. Weekends. Relentless. Methodically filtering, triaging, and eliminating, in record time, a long list of material issues and protocol-level improvements. Every fix is another layer of hardening ahead of Supernova.
To put things in perspective: in the last few months, we've surfaced and closed more issues than in the previous 3 - 4 years combined. This is the preparation phase. And the fact that AI has crossed a new capability threshold cannot be overstated.
Supernova is not an upgrade. It is a complete rewrite of what a blockchain can be, instant performance, radical scalability, human UX that actually feels seamless, and just works. It also opens a new verifiable agentic highway. Real. Live. Working today.
Continuously adaptive security will define the AI era—and almost no one is treating it that way.
Static firewalls defended the old internet. Continuously adaptive security is what defines survival in the AI era.
Our engineering team, along with a resourceful network of builders, validators, and community members, is doing the surgical, disciplined, vital work that shapes what comes next.
This is MultiversX. This is SUPERNOVA.