Chinezii acuzați că ”fură” tehnologia AI prin mii de conturi false şi reţele de pe dark web

Foreign companies are allegedly using fraudulent accounts and stolen credit cards to access advanced AI models from the U.S., according to Anthropic. This practice, known as "distillation," allows developers to train their own models using responses from existing AI systems. Anthropic claims that companies, particularly from China, are circumventing restrictions to illegally copy their Claude model. The company has accused Moonshot AI of using Claude without permission. Cybersecurity experts note that attempts to access these models are not limited to China, with other countries also involved. Anthropic warns that this issue poses national security risks, as malicious actors could exploit AI for surveillance or biological weapon programs. The company emphasizes the importance of protecting intellectual property while acknowledging the value of legal competition.