While Anthropic was still struggling with the leak of 512,000 lines of source code, this "engineering accident" unexpectedly became a "lifesaver" for domestic developers. According to cnBeta, due to the source code revealing Anthropic's extremely strict account review mechanism, a tool called CC-Gateway has emerged.
Previously, Anthropic was mocked by some developers as "A÷" (a homophone for "division sign") because of its "better to kill by mistake than to let one slip through" banning strategy towards IP addresses and accounts in China. This time, the "exposure" of the source code finally allowed the outside world to see how the company was monitoring users behind the scenes.
Monitoring Details are Shocking: A "Full Body Scan" Every 5 Seconds
The leaked code shows that Anthropic's risk control system is far more complex and radical than its peers. Its integrated telemetry and detection methods can be described as a "sky net":
More than 640 telemetry methods: from device ID, operating system configuration to browser fingerprint, covering everything in detail.
Frequency of once every 5 seconds: user information is reported in real-time at a very high frequency, and any small environmental fluctuation may trigger the account ban logic.
More than 40 types of anomaly detection: specifically targeting proxy environments and identity spoofing, performing multi-dimensional cross-validation.
In response to the exposed monitoring logic, developers quickly launched
Through this method, the data received by Anthropic's servers will become extremely uniform and normal, significantly reducing the probability of being automatically banned due to "environmental anomalies." However, industry insiders also warned that this "cat-and-mouse game" will continue as Anthropic continues to upgrade its telemetry system.
