Detect Ransomware by Scanning Text/HTML for Common Phrases
Could you possibly add the ability to scan newly created text or html documents for common phrases found in the ransom notes of current ransomware strains? Seems to be a common sense approach to detecting this kind of infection and preventing it from spreading much beyond the original point of infection. Think of it as DLP in reverse, people really shouldn't have the need to type things like "What happened to your files" or "Your files have been encrypted using the latest..." so the only possible source would be malicious software. On a Windows server you could easily get the owner of the file and change the ACL on all network shares to either block that user or change the rights to read only.
Jeff B commented
Found out today that your competition, NOD32, already detects this, better catch up.