Patents

United States patents where Adam Wosotowsky is listed as an inventor. Data pulled directly from Google Patents.

US11743276B2

Methods, systems, articles of manufacture and apparatus for producing generic IP reputation through cross-protocol analysis

Combining signals across different network protocols into a single, protocol-agnostic IP reputation score that is harder for attackers to poison than any single-source list.

Most IP reputation systems in production security look at one channel at a time: an IP is judged bad because it sent spam, or because it hosted phishing, or because it appeared in a malware sinkhole. Attackers exploit that silo — they burn the channel that got them flagged and pivot to a new one where their reputation is still clean. This work brings the channels together. Telemetry from separate protocols is normalized, correlated, and combined into a single reputation score that reflects everything the network is doing, not just one facet of it. In practice that means an IP has to earn a clean reputation across all the ways it shows up, not just the ones a defender happens to be watching, which raises the operational cost of the attacker's pivot. The impact for customers is fewer false positives (a benign mail server does not get slandered by an unrelated port scan) and fewer false negatives (a machine cleaning up its spam behavior while quietly hosting a C2 no longer looks clean). Underlying much of my published IP and domain reputation work at McAfee and Intel.

Assignee: Mcafee, Llc · Grant/Filing: 2023-08-29

US11689550B2

Methods and apparatus to analyze network traffic for malicious activity

Building a graph of related network entities from raw IP telemetry so that malicious behavior can be recognized by its structural shape, not by matching a specific attack signature.

Signature-based detection ages badly: as soon as a defender publishes what a campaign looks like, the campaign changes what it looks like. This work models network traffic as a graph — addresses, connections, and their relationships — and detects malicious activity from the structure that campaigns tend to share, not from any single indicator. The advantage is generalization: a new botnet or C2 network that looks different but behaves like its predecessors is still recognizable, because the shape of its infrastructure gives it away. In production this shows up as detections that survive attacker retooling, catch previously unseen variants, and give analysts a starting point (the graph itself) rather than just a verdict. This was one of the internal-tooling wins that made research iteration meaningfully faster on the team.

Assignee: Mcafee, Llc · Grant/Filing: 2023-06-27