The recent discovery of command-and-control software sending instructions to malware-infected computers from Dropbox raises the question of how such threats can be discovered.[Hackers found controlling malware and botnets from the cloud]Interviews with security consultants indicate that the new development in cloud-based malware can be detected by monitoring for particular anomalies in the network, since malware at some point acts differently than legitimate software.Vendor Trend Micro reported Thursday finding Dropbox-hosted C&C instructions for malware in botnets and compromised systems.The malicious activity was not the result of vulnerabilities in the file-sharing service. The cybercrooks simply opened up an account to start their criminal activity. The same operation could easily run on another cloud service provider.The easiest solution to threats from such services is to block employees from using them from the corporate network. As an alternative, companies could then build a similar service internally or provide employees with access to a single, more secure service provider."There should be no reason that you would want to have critical data, or the possibility of critical data, being accidentally or intentionally shared or put on a server that you do not have any security control over," Dave Chronister, co-founder and managing partner of Parameter Security, said.Fair enough. But companies can provide employees with more choices, if the organizations watch for activity that would indicate malicious code on a service provider talking to malware in the corporate network, Jonathan Thompson, chief executive of Rook Security, said.Indicators include:The opening of previously unused TCP\/IP ports, such as 22, 23, 80 and 8080, for data sharing. The Internet Relay Chat (IRC) system should also be watched.A computer starts reaching out consistently to other computers on the network, one after the other. "This behavior is more indicative of a compromised user computer, where the malware or attacker is using that compromised system to continue attacking other systems in the network," Thompson said.A system known only to be active from 8 a.m. to 4 p.m. suddenly has activity outside of that timeframe.A system that sends out 500 MB of data on an average day suddenly starts sending 5 GB of data after normal working hours.NSS Labs, which tests security products and sells its research on a subscription basis to corporations, recommends the use of a breach detection system."We found that most of the products offered very good malware callback detection, often higher than the detection of the actual malware file itself," Thomas Skybakmoen, an NSS Labs research director, said.However, there was one weakness with several of the products tested. "If the malware used proprietary protocols or SSL, it would increase the potential for evasion of these products," Skybakmoen said.[How to avoid having your cloud-hosted business destroyed by hackers]The products tested included AhnLab MDS, Fidelis XPS Direct 1000, FireEye Web MPS 4310 and Email MPS 5300, Fortinet FortiSandbox 3000D, Cisco's Sourcefire Advanced Malware Protection and Trend Micro Deep Discovery Inspector Model 1000.