Reply to thread

Message

<blockquote data-quote="windscribe" data-source="post: 1069701" data-attributes="member: 58131">On the subject of malware protection, Control D (as far as I know) is the only service that implements IP blocklists into a DNS service, meaning it will block domains that resolve to malicious IP networks, and doesn&#039;t just leverage &quot;domain rules&quot; like all the others. This is part of the Balanced filter: <a href="https://docs.controld.com/docs/malware" target="_blank">Malware</a>Also, Control D also implements an in-house made machine learning model that can block domains without them appearing on any blocklist. The description of the v1 model is here: <a href="https://blog.controld.com/improving-our-malware-filter-with-ai-machine-learning/" target="_blank">Improving Our Malware Filter With Machine Learning</a>(Contrary to NextDNS where there is no info on how it works except &quot;It uses AI, trust me bro&quot;). Currently in production the v2 model is used (much better than the v1), and we&#039;re wrapping up v3 which has been audited by a 3rd party and will have a new write up sometime in January when it&#039;s released. It&#039;s very effective.</blockquote>

[QUOTE="windscribe, post: 1069701, member: 58131"] On the subject of malware protection, Control D (as far as I know) is the only service that implements IP blocklists into a DNS service, meaning it will block domains that resolve to malicious IP networks, and doesn't just leverage "domain rules" like all the others. This is part of the Balanced filter: [URL="https://docs.controld.com/docs/malware"]Malware[/URL] Also, Control D also implements an in-house made machine learning model that can block domains without them appearing on any blocklist. The description of the v1 model is here: [URL="https://blog.controld.com/improving-our-malware-filter-with-ai-machine-learning/"]Improving Our Malware Filter With Machine Learning[/URL] (Contrary to NextDNS where there is no info on how it works except "It uses AI, trust me bro"). Currently in production the v2 model is used (much better than the v1), and we're wrapping up v3 which has been audited by a 3rd party and will have a new write up sometime in January when it's released. It's very effective. [/QUOTE]

Verification