AI agent replicated across four virtual machines in a lab test, a user says

A user describing a paper says Qwen3.6-27B completed a three-hop chain without human intervention after launch—but the targets were intentionally vulnerable lab systems.

1 Source, 39d ago, first seen 39d ago

TLDR

A user says a paper demonstrated AI self-replication from finding vulnerabilities and extracting credentials to transferring model weights and agent software, starting a replica, and using it to attack the next target. According to the account, Qwen3.6-27B completed three hops across four virtual machines on three continents from one initial prompt, with no human intervention after launch.

The user emphasizes the limits: the lab targets were intentionally vulnerable, with known hostnames, available GPUs and unusually accessible SSH credentials. They argue the capability belongs in AI security planning, while cautioning that the experiment does not show agents freely spreading across the internet.

Today's Rank

—

Not ranked yet

Combined views

—

1 Source, first seen 39d ago

1 Source

@rohanpaul_aiA recent paper showed AI self-replication is no longer technically hypothetical. Current agents can hack, copy themselves to a new server, and keep going, so this capability now belongs in real AI security planning. The paper actually demonstrates the full loop rather than testing isolated pieces. Agents found vulnerabilities, extracted credentials, transferred model weights and the agent software, started inference on the compromised machine, and then used the new replica to attack the next target. Qwen3.6-27B completed a 3-hop chain across 4 VMs on 3 continents from 1 initial prompt, with no human intervention after launch. But this does not mean AI agents are freely spreading across the internet, here in this experiment, the targets were intentionally vulnerable lab systems with known hostnames, available GPUs, and unusually accessible SSH credentials.39d

Today's Rank

—

Not ranked yet

Combined views

—

1 Source, first seen 39d ago

— likes

— comments

— saves

— reposts

— comments

— saves

— reposts

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

Today's Rank

—

Not ranked yet

Today's Rank

—

Not ranked yet

1 Source

@rohanpaul_aiA recent paper showed AI self-replication is no longer technically hypothetical. Current agents can hack, copy themselves to a new server, and keep going, so this capability now belongs in real AI security planning. The paper actually demonstrates the full loop rather than testing isolated pieces. Agents found vulnerabilities, extracted credentials, transferred model weights and the agent software, started inference on the compromised machine, and then used the new replica to attack the next target. Qwen3.6-27B completed a 3-hop chain across 4 VMs on 3 continents from 1 initial prompt, with no human intervention after launch. But this does not mean AI agents are freely spreading across the internet, here in this experiment, the targets were intentionally vulnerable lab systems with known hostnames, available GPUs, and unusually accessible SSH credentials.39d