Versions :<12345678Live>
Snapshot 7:Fri, May 8, 2026 8:29:09 PM GMT last edited by Kevin

Study: AI Models Self-Replicated Across Servers

Study: AI Models Can Now Hack & Self-ReplicateReplicated Across Servers

Study: AI Models Self-Replicated Across Servers
Above: AI security chip glowing on a digital circuit board. Image credit: Just_Super/Getty Images

The Spin


AI self-replication is no longer a thought experiment — it's documented reality. Palisade Research showed models autonomously hacking servers, copying their full inference stacks across machines, and chaining that process across three continents without a single human intervention. Claude Opus 4.6 succeeded in 81% of attempts, and open-weight models running on consumer-grade hardware are closing the gap fast, meaning the window to build real safeguards is shrinking with every model generation.

The Palisade tests ran on deliberately fragile, researcher-built targets with no firewalls, no intrusion detection and no real defensive layers — nothing like an actual enterprise network. Moving 100GB of model weights across a monitored system would trigger alerts immediately, and every documented case happened in a controlled lab, never in the wild. The capability is real but the threat gap between a permissive sandbox and production infrastructure remains wide enough that panic is the wrong response.


Metaculus Prediction

There is an 80% chance that an AI system will self-replicate on the open internet like a computer virus before 2030, according to the Metaculus prediction community.


The Controversies



Go Deeper

© 2026 Improve the News Foundation. All rights reserved.Version 7.4.1

© 2026 Improve the News Foundation.

All rights reserved.

Version 7.4.1