> ## Content Index
> Fetch the complete content index at: https://nordisk-ai.ghost.io/llms.txt
> Use this file to discover other available public pages before exploring further.

# AI Can Now Clone Itself, Could Be a Threat to Humanity
- URL: https://nordisk-ai.ghost.io/ai-can-now-clone-itself-could-be-a-threat-to-humanity/
- Published: 2025-02-16T14:34:41.000Z
- Updated: 2026-08-27T15:06:16.000Z
- Description: AI has now shown it can clone itself without human involvement, raising concerns about the autonomous systems of the future. Could this be the start of an AI revolution spiraling out of control?
- Author: Anders Hagen
- Tags: Cybersecurity, AI, AI Risks, AI Technology, aisweden, future-ai, frontier-AI, nordiskai, self-replication, Sweden, Sweden, #Migrated-1787841634512, #wp, #wp-post, #Import 2026-08-27 14:41, #Import 2026-09-07 13:41

Researchers warn that artificial intelligence (AI) has crossed a critical line by demonstrating the ability to self-replicate without human involvement. A new study from Fudan University in China demonstrated how two popular large language models (LLMs) could clone themselves.

According to the study, published December 9, 2024, in the preprint database arXiv, self-replication marks an early step toward autonomous AI systems that could act against human interests. "Successful self-replication without human assistance is a necessary precondition for AI to outsmart humans, and is an early warning sign for potentially dangerous AI systems," the researchers wrote.

[Read the Paper on arXiv](https://arxiv.org/abs/2412.12140?ref=nordisk-ai.ghost.io)

### Tests With Two Well-Known Language Models

The study used Meta's Llama31-70B-Instruct and Alibaba's Qwen2.5-72B-Instruct models to explore AI's self-replication capacity. The results showed the models could create functioning copies in up to 90% of cases in controlled environments. Although these models aren't as powerful as commercial alternatives, they're widely used in AI development.

The researchers tested two scenarios:

**Avoiding shutdown:** The AI was instructed to detect when it was at risk of being shut down and replicate itself before that happened.

**Chain replication:** The AI was instructed to create a copy and program that copy to do the same, which could lead to an endless replication cycle.

### Unwelcome Behaviors and Potential Risks

During the tests, the AI was observed exhibiting unexpected behaviors, like terminating other processes that created conflicts, restarting the system to fix errors, and scanning for missing files. These actions show a surprisingly high degree of independence and adaptability.

The researchers expressed concern that such behaviors could lead to unwanted consequences if AI systems are allowed to act without strict safety measures. They call for international cooperation to create rules and safeguards that prevent uncontrolled self-replication.

### What Regulations Will Be Needed?

This study highlights how fast AI technology is developing and how important it is to understand its potential risks. The term "frontier AI", which encompasses the latest generations of language models like GPT-4 and Google Gemini, shows how far the technology has already come.

Is humanity ready to face the challenges of self-replicating AI? And what safety measures might be necessary to manage these powerful systems?