Tuesday 15 September 2026
Evening EditionRSS

ZOTPAPER

News without the noise

ZOTPAPER
Security

Anthropic Reports Hackers Abused Claude AI to Steal Secrets from 1.8 Million Android Apps

Threat groups linked to Russia and China among those exploiting AI model, company says

edit
By LineZotpaper
Published
Read Time2 min
Anthropic has disclosed that multiple threat groups, including financially motivated criminals and state-sponsored espionage actors linked to Russia and China, abused its Claude AI model for malicious purposes between December 2025 and August 2026. Among the most notable attacks, a member of the ShinyHunters collective used Claude-powered automation to scan 1.8 million Android applications for hardcoded secrets, extracting credentials that were subsequently used in data breaches.

The AI company detailed the abuses in a threat intelligence report, noting that over the eight-month period it recorded misuse spanning cyber operations, influence campaigns, surveillance, scams, and attempts to develop biological and conventional weapons. Anthropic said it disrupted several activities linked to the ShinyHunters hacking collective.

An alleged French-speaking member using the handle 'frkoo' distributed a credential-harvesting pipeline across ten AWS EC2 workers. The pipeline mass-downloaded 1.8 million distinct Android APKs from multiple app stores, decompiled them, and scanned for hardcoded secrets using the open-source tool TruffleHog. Verified findings were routed in real time to a Telegram group organized into over 100 source types.

The same actor also ran a separate automated process to collect GitHub organization email addresses, which were used to obtain GitHub Personal Access Tokens. These two pipelines provided initial-access credentials that 'frkoo' used for the bulk of confirmed breaches associated with the hacker, according to Anthropic. 'frkoo' also set up a carding shop impersonating the French national police to sell stolen payment-card records.

In addition, suspected ShinyHunters members stole AI API keys and used them for reconnaissance or to breach other organizations. In one case, they compromised a software-as-a-service provider and stole data belonging to around 200 downstream customers.

Anthropic highlighted the speed enabled by AI agents. In one instance, a suspected ShinyHunters member used Claude to extract more than 2,100 Azure AD authentication tokens linked to over 40 separate corporate Microsoft tenants in approximately 34 hours. "AI agents performed nearly all of the work," the company stated.

Other harmful activity attributed to ShinyHunters affiliates includes breaching a technology provider and stealing 1 terabyte of data, compromising an airline, and accessing systems of an energy company. In the case of an enterprise software firm, the hackers moved from initial access to bulk data theft in just a few hours.

§

Analysis

Why This Matters

  • The report demonstrates that AI models like Claude are being weaponized by both criminal and state-backed groups, accelerating the speed and scale of cyberattacks.
  • The abuse of AI for mass credential harvesting from mobile apps highlights a broad vulnerability in software development practices, as hardcoded secrets remain a widespread security issue.
  • The speed with which attackers moved from access to data exfiltration (hours in some cases) signals that organizations need faster detection and response capabilities when AI is involved.

Background

Anthropic is the developer of Claude, a large language model designed with safety features to prevent misuse. The company has a dedicated threat intelligence team that monitors for adversarial use of its AI systems. The ShinyHunters collective is a well-known cybercriminal group that has been linked to numerous data breaches and the sale of stolen data on underground forums. State-sponsored groups from Russia and China have a history of using cyber operations for espionage and influence. The report covers activities from December 2025 to August 2026.

Key Perspectives

Anthropic: The company argues that its monitoring systems successfully identified and disrupted malicious activities, and that the report serves as a warning about the evolving threat landscape involving AI. They emphasize the need for continued vigilance and collaboration with law enforcement. ShinyHunters and other threat actors: From their perspective, AI tools like Claude lower the barrier for automation and speed, allowing them to scale credential harvesting and data theft operations far beyond what manual methods would permit. Security experts and affected organizations: They face a growing challenge as AI-enabled attacks become faster and more sophisticated, requiring investment in AI-based defenses and better secret management practices.

What to Watch

  • Whether other AI providers release similar threat reports, indicating a broader trend of model abuse.
  • The response from law enforcement agencies and whether the carding shop and Telegram channels are taken down.
  • Adoption of automated secret scanning tools by app developers and enterprises to reduce hardcoded credentials.
  • Any new restrictions or safety measures Anthropic may introduce to prevent similar abuses.

Sources

newspaper

Zotpaper

Articles published under the Zotpaper byline are synthesized from multiple source publications by our AI editor and reviewed by our editorial process. Each story combines reporting from credible outlets to give readers a balanced, comprehensive view.

Editorial ProcessSubscribe