Also Windows Defender might have options to do live cloud verification or other levels of threat verification like generic heuristics. Is the web connection enabled in the VM and all Windows Defender settings the same? Virustotal style hash checking and stuff are becoming more common in antivirus apps lately for having access to a more up to date and broader database that allows vendors to find viruses earlier as well. Could even be some random spyware setting in your Windows account profile usually under the title of "help Microsoft improve our products and user experience" type of option.
Or Windows Defender is so smart that it knows when you are in a VM or sandbox probably you are studying the viruses and do not want to block them. But doubt it