I wonder why they would use something so inefficient in this context. Are there phishing scanners that look for more efficient (high-entropy) strings of characters, like base64?