A security researcher conducted AI-assisted reverse engineering on TP-Link Tapo C200 cameras and discovered multiple vulnerabilities affecting 25,000 internet-exposed devices. The researcher obtained firmware from an open S3 bucket, decrypted it using existing tools, and used AI to analyze the code in Ghidra. They found vulnerabilities including hardcoded SSL private keys enabling traffic decryption, pre-authentication memory overflow in ONVIF SOAP XML parsing, HTTPS Content-Length integer overflow causing crashes, and authentication bypass allowing device hijacking through malformed JSON requests.