---
title: "Anthropic Accuses Alibaba of Cloning Claude Using 25,000 Fake Accounts: What We Know"
url: https://digitaltechbyte.com/anthropic-alleges-alibaba-cloned-claude-using-25000-fake-accounts/
date: 2026-06-27
modified: 2026-06-27
author: "Brijesh Desai"
description: "Anthropic Accuses Alibaba of Cloning Claude using 25,000 fake accounts and millions of Claude AI interactions in a suspected model extraction campaign. Here’s the latest update. Anthropic alleges Alibaba cloned..."
categories:
  - "News"
tags:
  - "agentic reasoning"
  - "AI model theft"
  - "AI security"
  - "Alibaba Qwen"
  - "Anthropic"
  - "Claude AI"
  - "fake accounts"
  - "large language models"
  - "model distillation"
  - "software engineering"
image: https://digitaltechbyte.com/wpbytes/wp-content/uploads/2026/01/anthropic-claude-1024x536.webp
word_count: 631
---

# Anthropic Accuses Alibaba of Cloning Claude Using 25,000 Fake Accounts: What We Know

Anthropic Accuses Alibaba of Cloning Claude using 25,000 fake accounts and millions of Claude AI interactions in a suspected model extraction campaign. Here’s the latest update.

# Anthropic alleges Alibaba cloned Claude using 25,000 fake accounts

Anthropic alleges Alibaba cloned Claude using 25,000 fake accounts in what the company describes as the largest known effort to extract its AI model’s capabilities. The claim has quickly become one of the most watched AI security stories of the week because it sits at the intersection of model theft, U.S.-China tech competition, and the growing pressure on frontier AI labs to protect their systems.

Anthropic says the activity was not a one-off misuse but a coordinated distillation campaign that ran across thousands of fraudulent accounts and millions of interactions. The allegation has added fresh urgency to a debate the AI industry has been having for months: how easy is it for one lab to study another lab’s model behavior and recreate much of its value at lower cost?

## What Anthropic claims

According to reporting on the company’s letter to U.S. lawmakers, Anthropic says operators linked to Alibaba and its AI lab Qwen conducted nearly 29 million exchanges with Claude between April 22 and June 5 using about 25,000 fraudulent accounts. The company says those interactions were aimed at pulling out Claude’s capabilities rather than simply using the chatbot normally.

Anthropic describes the alleged tactic as “distillation,” a method in which a smaller or cheaper model is trained on outputs from a stronger model to imitate its performance. In its telling, the effort targeted especially valuable abilities such as software engineering, agentic reasoning, and long-horizon task handling.

## Why this case stands out

What makes this allegation unusual is scale. Anthropic says this was the largest known attack of its kind against the company, and one of the biggest efforts yet by a Chinese AI player to reverse-engineer a leading U.S. model. The reported use of fake accounts and proxy-like obfuscation also suggests the campaign may have been designed to avoid easy detection.

This matters because modern AI models are increasingly judged not only by raw benchmark scores but by practical usefulness in coding, reasoning, and workflow automation. If a rival can repeatedly probe a model and collect enough output patterns, it may be able to narrow the performance gap without building everything from scratch.

## The broader AI backdrop

The allegation lands at a tense moment for the global AI sector. Companies are racing to release more capable systems, while governments are paying closer attention to model security, data governance, and intellectual property protection. Anthropic has also previously said it has seen similar extraction attempts involving other Chinese AI firms, which suggests this is part of a broader and ongoing pattern rather than a single isolated episode.

For AI developers, the issue is bigger than one dispute. It raises questions about how frontier models are accessed, how abuse is measured, and what technical barriers can realistically stop large-scale scraping or probing when attackers can spread activity across thousands of identities.

## What happens next

So far, these remain allegations reported by Anthropic and media outlets, not a court ruling or public admission from Alibaba in the sources reviewed. Anthropic has reportedly urged lawmakers to take the issue seriously, which could push the story from a technical dispute into a policy fight over AI model protection and competition rules.

For readers, the key takeaway is simple: this is not just a headline about one chatbot. It is another sign that the AI race is moving into a phase where protecting model capabilities is becoming as important as building them.

**Summary:** Anthropic says Alibaba-linked operators used about 25,000 fake accounts and millions of Claude interactions in a suspected model-extraction campaign, making this one of the biggest AI cloning allegations yet.