At least in terms of malware detection, Chrome utilises a bloom filter in the first instance to identify the probability of a URL being malicious before making remote calls. If it is found to be positive, only then does it submit it to Google for more precise verification.