فضيحة الشفرة المظلمة.. كيف اخترقت إيران Claude وحولته إلى جاسوس؟

In one of the most terrifying technical paradoxes in Silicon Valley, the company "Anthropic" revealed a sophisticated security breach that affected its advanced model Claude, after entities linked to the Iranian regime successfully harnessed artificial intelligence to serve political espionage and direct digital surveillance objectives.

The operation was not just about traditional inquiries; rather, the intelligent model transformed into an "intelligence brain" that analyzes scattered data and accurately identifies individuals.

Claude the Digital Spy

According to the official report issued by the company, Iranian entities developed a malicious mechanism that relies on processing Claude's analytical capabilities, as the model was fed with personal accounts and images of posts on social media platforms.

Claude connects these scattered data to match identities, determining the real names and locations of Iranian opponents and members of minorities abroad, who are typically targeted by the regime using traditional intimidation methods.

The scandal did not stop at Iranian borders; investigations revealed that the exploitation activities of Claude extended to:

• Dangerous biological research: Attempts to conduct research on highly contagious viruses such as "bird flu" and "monkeypox" and deadly toxins.
• Exploitation in West Africa: A contractor with the national security agencies in Mali used the model to design software for gathering intelligence.
• Data theft in China: Prominent Chinese companies secretly exploited Claude's responses to "distill" them and train their own models.

Upon discovering these breaches, the company closed the involved accounts and banned the processing mechanisms, warning that the absence of legal controls makes artificial intelligence tools a terrifying weapon that could turn into the largest espionage net in human history.