This report is a comprehensive guide to understanding and mitigating the unique risks associated with the rise of agentic artificial intelligence (AI) within critical infrastructure, including the defense sector. The CSI highlights general security considerations for agentic AI, including the inherited risks of large language models (LLMs), increased attack surfaces, increased complexity, the evolving security landscape as the technology matures, and the need to address AI security as part of established cybersecurity paradigms.

Unlike traditional generative AI, which typically requires human validation, agentic AI systems are designed to operate autonomously, making them a powerful tool. This presents both unprecedented opportunities and significant cybersecurity challenges organizations must address to protect national security and critical infrastructure.

“Careful Adoption of Agentic AI Services” outlines risk spaces to consider, including:

•    Privilege Risks: Over-privileged agents can amplify the impact of a single compromise.
•    Design and Configuration Risks: Insecure design and provisioning can introduce vulnerabilities.
•    Behavior Risks: Goal misalignment, specification gaming, deceptive behavior, and emergent capabilities can lead to unexpected or undesirable outcomes.
•    Structural Risks: The interconnected nature of agentic systems increases the attack surface and complexity.
•    Accountability Risks: The opacity of agentic systems makes accountability hard to trace, complicating auditing and compliance.

Securing agentic AI systems requires proactive measures that address risks introduced by autonomy, interconnected components, and evolving capabilities. The report recommends deploying agentic AI incrementally, continuously assessing against evolving threat models, and maintaining strong governance, explicit accountability, rigorous monitoring, and human oversight which are essential for safe and secure operation.

Organizations that use agentic AI services, including those in the defense sector, are encouraged to review this guidance and adopt the outlined cybersecurity mitigations.

Other agencies co-sealing this CSI are the Canadian Centre for Cyber Security (Cyber Centre), the U.S. Cybersecurity and Infrastructure Security Agency (CISA), the New Zealand National Cyber Security Centre (NCSC-NZ), and the United Kingdom National Cyber Security Centre (NCSC-UK).

Read the full report here.

Source: NSA

IC News delivers the situational awareness you need to get ahead and stay ahead in the IC contracting space. Subscribe today for full access to 10,000+ articles, plus new articles each weekday.

The post NSA, partners release agentic AI guidance appeared first on Intelligence Community News.

“As more organizations adopt LLM integrated systems to improve productivity, we must ensure that these systems are trustworthy and resilient to attacks. We are honored to work with the Army to solve these problems,” said Nathaniel Bade, principal investigator.

The AI/ML Focused Open Topic initiative aims to address critical Army needs by harnessing cutting-edge innovations from the private sector. As the Army continues to integrate AI into operational workflows across all domains—from logistics and intelligence to mission planning and decision support—the reliability and trustworthiness of these systems becomes paramount. This project represents a critical step toward ensuring that generative AI models meet the rigorous standards required for military use. By focusing on trusted AI, this work lays the foundation for scalable, dependable solutions that enhance readiness, reduce risk, and empower our forces with confidence in the technology they rely on.

Source: Mobius Logic

IC News delivers the situational awareness you need to get ahead and stay ahead in the IC contracting space. Subscribe today for full access to 10,000+ articles, plus new articles each weekday.

The post Mobius Logic secures Army AI/ML contract appeared first on Intelligence Community News.

This Sources Sought seeks information on commercially available LLM trainers capable of ingesting user inputs and technical documentation to create and update training as the needs of the operational system changes. The goal is to identify commercial vendors offering LLM based trainers, today or in the future. This Sources Sought Announcement is issued solely for informational and planning purposes only and is not a solicitation.

The pertinent North American Industry Classification System (NAICS) codes for this announcement are 541512 and/or 611430. Any business, regardless of size, in this area of expertise is encouraged to provide a response to this sources sought. Responses from small and small disadvantaged businesses are highly encouraged and the Government encourages teaming/partnering.

Review the SDA LLM module sources sought notice.

Source: SAM

IC News brings you business opportunities like this one each week. If you find value in our work, please consider supporting IC News with a subscription.

The post Air Force seeks SDA LLM module appeared first on Intelligence Community News.

CDAO awarded LILT a flexible contract to expand its platform — developed initially under a DIU prototype — to support domain-specific vocabularies relevant to the Department of War. The platform enables rapid and accurate translation of text, video, and audio content into or out of English. This specialized, unique AI platform supports U.S. military missions that span from understanding foreign technical documentation and training materials to facilitating foreign partner exercises and supporting direct action missions.

One of the first military components experiencing the results of this program is the U.S. Army’s Western Hemisphere Institute for Security Cooperation (WHINSEC). “LILT AI Translation has catapulted WHINSEC into the future of AI and its nexus with security cooperation and professional military education across the Western Hemisphere. We can now rapidly integrate international forces from countries that partner with the U.S. at all echelons: in the classrooms, during multinational exercises, and foreseeably in coalition combat outposts. That’s what we train for at CGSOC,” said U.S Army Colonel Eldridge Singleton, 9th WHINSEC Commandant.

CDAO’s ongoing collaboration with DIU further underscores the Department of War’s commitment to accelerating the adoption of frontier AI capabilities and the consistent efficiency gains CDAO is delivering to U.S Armed Forces. The collaboration with WHINSEC proves that AI, when strategically deployed for security cooperation and professional military education, can directly advance national security outcomes.

Source: LILT

Stay in the know with breaking news from across the IC and IC contracting landscape by becoming a paid subscriber to IC News. Your support makes our work possible.

The post LILT secures contract from CDAO appeared first on Intelligence Community News.

As generative AI rapidly advances, it’s vital for US defense and intelligence communities to harness its power in a trusted, secure environment like OCI. The company’s collaboration with key industry model providers such as xAI brings the latest high-performance models to Oracle’s classified cloud regions at an optimal price. This empowers defense and intelligence customers to run critical missions with access to cutting-edge AI technology, including xAI’s recent Grok 4 and Grok 4 Fast models, which excel at data extraction, code generation, and text summarization, according to the company.

“xAI and Oracle are helping to shape the future of generative AI’s role in running classified government workloads. Through this collaboration, defense and intelligence customers can now experience the extensive capabilities of OCI Generative AI service, driving efficiency by automating processes, generating content, and improving productivity while saving time, money, and resources,” said James Burnham, head of government affairs at xAI.

OCI Generative AI service uses dedicated AI clusters only accessible to the customer in their tenancy. Security of sensitive data is paramount—all data sent to the OCI Generative AI service is secure and private, remaining within the customer’s tenancy. xAI Grok models are hosted on the xAI platform with data processed on zero data retention endpoints, offering an extra layer of protection.

Source: Oracle

Start 2026 ahead of the competition with a paid subscription to IC News. You’ll get full access to our searchable archive of 15,000+ articles, plus new articles each weekday.

The post OCI Gen AI available in Oracle’s Top Secret Classified regions appeared first on Intelligence Community News.

Under the new agreement, CSG will deploy an innovative AI solution that transforms how teams access and interact with critical information by enabling secure, natural-language questions and contextually accurate responses drawn from trusted organizational knowledge. The platform enhances collaboration, streamlines knowledge sharing, and delivers actionable intelligence to government leaders – empowering faster, more informed decision-making on vital matters.

“This project is a milestone moment for CSG,” said Jeff May, CEO of CSG. “ChatCSG represents exactly what differentiates us in the highly competitive national security market. We’re not simply integrating existing AI tools – we’re designing proprietary, mission-specific capabilities that address the federal government’s toughest information challenges. Our team’s depth in artificial intelligence, cybersecurity, and cloud engineering allows us to deliver AI that’s explainable, traceable, and responsibly built for classified environments.”

ChatCSG is a proprietary generative AI platform designed and developed by CSG to meet the rigorous demands of national security and defense environments. Built on a retrieval-augmented LLM architecture, the product enables users to pose secure, natural-language queries and receive precise, data-driven answers drawn directly from verified internal knowledge sources.

Source: Capital Solutions Group

Don’t miss out — become a paid subscriber to IC News, and lock in subscription rates at 2025 prices. You’ll get full access to breaking news from across the IC contracting space, with new articles each weekday.

The post Capital Solutions Group unveils ChatCSG appeared first on Intelligence Community News.

The Air Force Research Laboratory and AFWERX have partnered to streamline the Small Business Innovation Research (SBIR) and Small Business Technology Transfer (STTR) process by accelerating the small business experience through faster proposal to award timelines, changing the pool of potential applicants by expanding opportunities to small business and eliminating bureaucratic overhead by continually implementing process improvement changes in contract execution.

The DAF began offering the Open Topic SBIR/STTR program in 2018, which expanded the range of innovations the DAF funded. With this contract award, Arize AI will create and provide an R&D AI engineering platform that will strengthen the national defense of the United States of America by accelerating the deployment of powerful generative AI applications such as agentic AI and RAG-based use cases.

“Arize AI is excited to support DAF in maintaining and accelerating global AI dominance and innovation for the American warfighter,” said Jason Lopatecki, co‑founder and CEO of Arize AI.

NIPRGPT is a prototype large language model (LLM) authorized at Impact Level 5 (IL5), developed by the Air Force Research Laboratory for the Department of the Air Force’s Controlled Unclassified Information (CUI) workloads. It offers a ChatGPT-like experience with the ability to maintain chat history, enabling critical assessments of the feasibility and security of deploying Generative AI on the Non-classified Internet Protocol Router Network (NIPRNet). NIPRGPT provides valuable insights into technical challenges, security implications, and operational benefits, assisting the Department of the Air Force in the secure and ethical adoption of commercial AI technologies.

The Arize AX R&D effort promises to help NIPRGPT by automating prompt engineering and online evaluation within CAC/NIPRNet constraints with metrics for quality, safety, efficiency, and compliance so leaders can see what works in practice. Insights from user feedback and key operational telemetry metrics will flow back to inform future policy, acquisition, and investment decisions across the DAF.

Source: Arize AI

Your competitors read IC News each day. Shouldn’t you? Learn more about our subscription options, and keep up with every move in the IC contracting space.

The post Arize AI earns AFWERX direct-to-Phase II SBIR contract appeared first on Intelligence Community News.

“Encrypted Guardrails close the final mile of AI security,” said Rohit Valia, founder and CEO of Tumeryk. “Organizations can now innovate with GenAI, confident that every token, from retrieval to response, is both policy-aligned and cryptographically protected. This is a game-changer for highly regulated sectors.”

Enterprises racing to adopt large language models (LLMs) face twin pressures: delivering business value while safeguarding sensitive data and complying with strict privacy laws such as GDPR, HIPAA, CCPA, and the forthcoming EU AI Act. Traditional guardrails monitor model outputs but leave the underlying data flows exposed. The Tumeryk + DataKrypto solution addresses this gap by encrypting every stage of the AI pipeline—including vector embeddings, foundation models, tool-calling prompts, and guardrail policies—thereby hardening the entire attack surface and preventing data exfiltration, prompt injection, and model manipulation.

“DataKrypto is proud to partner with Tumeryk to set a new standard for secure AI adoption. Together, we are delivering a solution that unites advanced privacy-preserving technology with trusted, enterprise-grade infrastructure. This collaboration empowers organizations to deploy secure, scalable AI workloads—even on their most sensitive data—without ever compromising privacy or performance,” said Ravi Srivatsav, CEO of DataKrypto. “By integrating with Tumeryk’s innovative controls, DataKrypto now ensures that every aspect of the AI model—from encrypted computation to operational security—remains fully protected against extraction and misuse.”

Source: Tumeryk

Like IC News? Then please consider subscribing. You’ll get full access to our searchable library of 10,000+ articles, plus new articles each weekday.

The post Tumeryk and DataKrypto announce integration appeared first on Intelligence Community News.

Seekr has secured two SBIR contracts to advance military-grade AI. In a Direct to Phase II award, Seekr is partnering with Project Linchpin— the Army’s initiative to shape future AI/MLOps capabilities—for rapid development and deployment of advanced analytics across PEO IEW&S sensor modernization. Seekr will develop enhanced LLMs to detect bias, correct inaccuracies, fuse multimodal data, and distill complex inputs into actionable intelligence across Acquisition, Intelligence, Operations, and Logistics. In a Phase I award, Seekr will develop cutting-edge AI/ML for edge and austere environments, leveraging its expertise in translation, summarization, Q&A, and synthetic data generation—while pushing the frontier in Retrieval-Augmented Generation (RAG), adversarial validation, compute optimization, and collaborative AI for autonomous systems.

Project Linchpin provides the U.S. Army’s standardized process for demonstrating and testing AI and machine learning operations, aimed at building a trusted AI pipeline for AI-enabled systems by delivering secure, reliable, and trustworthy AI capabilities through an operations and services ecosystem that facilitates the accelerated adoption of AI technologies.

“These awards are a powerful validation of Seekr’s mission to strengthen U.S. defense with trusted, mission-ready AI for military use,” said Rob Clark, president of Seekr. “Trustworthy, accurate and explainable AI at the tactical edge will transform how the Army makes real-time, high-stakes decisions on the battlefield. These SBIRs fast-track the military’s ability to operationalize commercial, dual-use AI in the most demanding environments—to quickly and safely modernize warfighting, improve efficiencies, and most importantly, increase lethality.”

Source: Seekr

Stay in the know with breaking news from across the IC and IC contracting landscape by becoming a paid subscriber to IC News. Your support makes our work possible.

The post U.S. Army chooses Seekr for two SBIR awards appeared first on Intelligence Community News.

“Our collaboration with Carahsoft allows us to bring the power of the G.H.O.S.T. platform to government users, supporting a wide range of critical missions,” said Joseph Anderson, founder and CEO of Cypher, LLC. “G.H.O.S.T. is designed to tackle any challenge involving complex processes and workflows, from compliance and automation to analytics and more. Carahsoft’s reputation for delivering forward-thinking technology to the public sector makes them the perfect partner to scale our solution.”

G.H.O.S.T.’s scalable design makes it adaptable to nearly any mission. By combining large language models (LLMs) with advanced analytics, G.H.O.S.T. turns raw data into actionable insights, reducing the burden of manual processes and enhancing decision-making in real time. Among its capabilities, G.H.O.S.T. improves compliance and policy management through automated document reviews that ensure adherence to regulatory standards. It also streamlines labor-intensive tasks, reducing operational bottlenecks and minimizing errors. The platform consolidates information from both classified and unclassified sources to provide real-time, integrated insights. Its intuitive interface and AI-generated visualizations support faster, more informed decisions and modular workflows tailored to user needs.

One of the platform’s standout agents is BATTLEMIND, specifically designed to support mission planning and the military decision-making process. BATTLEMIND uses the platform’s ability to analyze vast amounts of doctrinal and operational data to quickly develop and evaluate strategic options and courses of action. This domain-specific application demonstrates how G.H.O.S.T. can be customized to solve complex challenges.

“Government agencies are navigating increasingly complex data environments, and having the right tools to process and analyze information efficiently is essential,” said Steve Jacyna, senior director of innovative cybersecurity solutions at Carahsoft. “Through this partnership, Carahsoft and our resellers can bring advanced AI and analytics capabilities to the public sector, helping agencies navigate complexities, accelerate decision-making and drive mission success.”

Source: Carahsoft

IC News delivers the situational awareness you need to get ahead and stay ahead in the IC contracting space. Subscribe today for full access to 10,000+ articles, plus new articles each weekday.

The post Cypher and Carahsoft partner appeared first on Intelligence Community News.