According to CoinMeta, three former security researchers warn that advanced AI models are becoming increasingly difficult to monitor, making it challenging for independent experts to identify risks and verify the security of these systems. This concern also affects companies that implement autonomous AI systems. As companies grant more power to these systems, they need to understand how they make decisions. Otherwise, it will be very difficult to trust these systems in performing important tasks. In an open letter released on October 8, 2026, the three researchers challenged their layoffs and denied leaking any information about the difficult-to-monitor AI architecture. OpenAI denied the allegations of retaliation and stated that they would not fire the researchers due to their concerns. The researchers emphasized the importance of cooperation with external experts and demanded that OpenAI maintain the accessibility of their models for independent security verification.