Leidos, a provider of technology solutions across aviation, defense, civil and health sectors, has advanced its cybersecurity capabilities with the $300 million acquisition of Kudu Dynamics, a specialist in offensive and defensive cyber operations.
The transaction, which was completed in May 2025, aligns with Leidos’ broader strategy to expand its national security technology portfolio through acquisitions and partnerships, reflecting the growing demand for advanced cyber solutions amid an increasingly complex threat environment.
Technology investment thru NorthStar 2030 strategy
NorthStar 2030 strategy is Leidos’ vision to improve threat detection and operational resilience via AI automation. The company, through DARPA military contracts, is developing AI-driven tools designed to identify and remediate cybersecurity vulnerabilities.
Leidos’ $120 million NATO cybersecurity contract highlights its growing role in protecting critical digital infrastructure. Through its NorthStar 2030 strategy, the company is advancing toward a more integrated, adaptive mode by investing in cloud, cyber, and AI capabilities while aligning growth with evolving customer and security demands.
Leidos: Patenting activity
From 2013 to 2016, the company sustained high levels of patent activity, reflecting a concentrated period of research and development. In 2016-2017, the company diverted its focus on the acquisition and integration of Information Systems & Global Solutions (IS&GS) business which coincided with the patent activity decline. Patent filing rebounded in 2018, highlighted even more by a key innovation award and the launch of Leidos Alliance Partner Network. This alliance is a strategic partnership at corporate and technological level with Dell, McAfee, Amazon Web Services, NetApp, Hewlett Packard Enterprise, Cisco, and Juniper.
In terms of national security and cybersecurity, Leidos continued its expansion by acquiring Dynetics in 2019 and L3Harris Technologies’ security detection and automation business in 2020, further strengthening its cybersecurity and automation portfolio.
This foundation of innovation supports Leidos’ ongoing efforts to strengthen its cybersecurity capabilities. The company has released an open-source AI tool to address evolving cyber threats and is investing in AI-enabled platforms for vulnerability remediation. In addition, Leidos secured a $120 million federal cybersecurity contract focused on improving compliance and IT efficiency, helping government clients better manage emerging cyber risks.
Leidos: Top Jurisdictions
Leidos has built a strong international IP portfolio, with the highest number of filings in the United States, followed closely by Canada and Australia.
Filings in Japan, India and Korea point to a growing strategic focus on the Asia-Pacific region, where Leidos appears to be positioning itself within key emerging innovation ecosystems. Patent coverage in Europe and the United Kingdom further indicates a global approach to intellectual property, aimed at ensuring long-term value and relevance across both established and evolving technology hubs.
Leidos: Top Law Firms
The company’s patent activity is supported by a diverse set of legal representatives across geographies. Leading the list is Bey & Cotropia, PLLC, reflecting Leidos’ strong focus on the U.S. market. Included on this list in the US market are McCarter & English LLP and Banner & Witcoff, Ltd.
Agencies such as RnB IP (Australia) and Cassan Maclean IP Agency (Canada) further demonstrate Leidos’ reliance on experienced regional partners to manage high volumes of patent applications.
Leidos: Top Technology Areas
Leidos’ patent filings are heavily concentrated in two technology areas. These are technologies under Heterocyclic compounds (C07D) and Preparations for medical/dental/toilet purposes (A61K), which makes up 50.7% of its portfolio. Related to this are peptides (organic compounds) (C07K), microorganisms or enzymes; compositions (C12N), fermentation or enzyme-based synthesis processes (C12P), and specific therapeutic activity of chemical compounds or medicinal preparations (A61P) which is related to Leidos’ push for the pharmaceutical and medical industry innovation.
Areas that focus on AI, cybersecurity, and defense systems are image processing (G06T), followed by electric discharge tubes or lamp apparatus (H01J). Significant portions cover areas such as analysis/investigation of materials (general) (G01N), and electric digital data processing (general computing) (G06F),
Key patents supporting Leidos’ cybersecurity innovations
The following patents highlight how Leidos is building a comprehensive cyber infrastructure, with developments in threat detection, security model training, and privacy-preserving AI. Each patent reflects the company’s strategic focus on strengthening national defense and advancing cybersecurity innovation in the coming decade.
Improved threat detection in computer systems
Illicit activities such as malwares and cyber-attacks have begun to become increasingly adaptive and difficult to detect, making existing computer software inadequate in inspecting evasive malware and differentiating legitimate and malicious acts. The emergence of various malicious activities have caused a great challenge to cybersecurity systems, as it becomes progressively harder to ensure security, confidentiality, and integrity of data in computer networks.
Kudu Dynamics’ U.S. Patent No. 10,291,645, which reflects a core aspect of the company’s innovations, addresses this challenge by developing a computer-based system that incorporates an anomaly detection framework. This system can detect illegal activities via a calculated metric which evaluates the probability of a computer activity being malignant, instead of relying on the conventional binary classification between benign and malicious activities. It first analyzes the behavior and context of computer activities and assigns a score to it relative to its likelihood of being malicious. This framework thereby improves the capacity of computer systems to automatically detect malwares, generate codes to exploit the vulnerabilities, and then obfuscate them to evade signature-based detection while blending in with normal computer operations.
The patent titled “Determining maliciousness in computer networks”, assigned under Kudu Dynamics, lists Mike Frantzen, Andrew Aarne Hendela, and Bryan Hafen Leavitt as inventors. It was filed on July 9, 2018, and was granted on May 14, 2019. They were represented by Chad Tillman, James Wright, Neal Wolgin, and David Higgins from Tillman Wright, PLLC.
Enhancing data augmentation for security-based machine learning models
Machine learning models are widely used in security-focused applications such as content filtering, malicious behavior detection, and network traffic classification. However, there is a disparity in the available training data, benign activity data is abundant, while malicious activity data is limited due to the infrequent and varied nature of computer attacks. While dataset augmentation can help mitigate this imbalance by transforming and manipulating existing data, these techniques are ineffective in cybersecurity cases since even the slightest alterations can distort the true classification of the data.
To address this, U.S. Pat. App. No. 2024/0248984 proposes a process that enhances the augmentation of cybersecurity data, tailored specifically for training machine learning models used in security tasks. The approach allows the generation of new training data without altering its original classification, whether the activity is benign or malicious, which preserves its intended behavior for model training. This helps expand the training dataset while also keeping the semantic integrity and accuracy of the data, which in turn improves the performance and reliability of the security-related machine learning models.
The patent application titled, “Process for generating offensive and defense security dataset augmentation with invariance and distribution independence,” was filed on January 22, 2024 and was published on July 25, 2024. The inventors are Paul Roysdon and Gavin Black, and were represented by Michael Jaro, Dawn-Marie Bey and Christopher Cotropia from Bey & Cotropia, PLLC.
Privacy-preserving federated learning
As part of the ongoing technological advancements, federated learning models have been implemented in computer networks to protect client data privacy. These models enable multiple users to collaboratively train machine learning models without directly sharing raw data. However, despite this setup, privacy risks still remain in the technical field. Existing solutions, such as aggregation techniques, aim to prevent unauthorized data exposure, yet, they often compromise model performance and accuracy, making users remain vulnerable to privacy attacks on the internet.
To address the issue of client privacy, U.S. Pat. App. No. 2024/0413969 introduces an upgraded federated learning framework which primarily secures client privacy upon use without compromising training performance. Unlike traditional federated learning frameworks, the invention employs a segmented model structure that only shares selected parts of the model from one client as requested. This approach also enables homomorphic encryption, a method that allows direct computation without raw data exposure which further protects client privacy. As a result, the system significantly reduces the risk of data leakage while maintaining high model accuracy and efficiency.
The patent application titled, “System and process for securing client data during federated learning,” was filed on June 12, 2024 and was published on December 12, 2024. Inventors listed are Olivia Galliker d’Aliberti, and Evan Michael Gronberg. The application was represented by Michael Jaro, Dawn-Marie Bey and Christopher Cotropia from Bey & Cotropia, PLLC.
