According to CoinMeta, David Sacks posted on platform X, criticizing Anthropic's latest research report for exaggerating the security risks associated with small AI models autonomously replicating and spreading on IoT devices. He pointed out that the experiments in question relied on devices with open debugging ports and manually set propagation tasks, which are not sufficient to prove the existence of AI worms that can spread autonomously. Sacks noted that traditional malware has already been able to spread on a large scale by exploiting vulnerabilities such as default device passwords and open ports, and small models have not significantly changed this threat pattern. He mentioned that Coinbase recently fine-tuned the Qwen-3.5-9B model, which performed better than Anthropic Opus 4.5, reducing fraudulent transactions by about 30%. Sacks believes that small models optimized for specific corporate tasks are competitive in terms of cost, speed, and effectiveness, and are posing a challenge to the business models of closed-source large model companies like Anthropic. They also question Anthropic's emphasis on the security risks of small models when these models gain market share.