IARPA Launches Five AI Research Programs for Intelligence Operations

The Intelligence Advanced Research Projects Activity has launched five new artificial intelligence research programs designed to accelerate the transition of emerging technologies into operational intelligence capabilities. The announcement comes through IARPA's Emerging Technology Accelerator framework, which uses other transaction authority agreements to streamline engagement with both traditional and non-traditional contractors.

According to the agency's official announcement, the five programs—ARCADE, COSMIC, DECIPHER, LocUS, and MOVES—represent a concerted effort to extract actionable insights from complex data sources including geospatial imagery, circuit design, linguistic trends, and open-source video content. IARPA's official newsroom release confirms the programs were initially introduced in January at an IARPA-hosted Proposers' Day attended by more than 550 researchers and industry representatives.

Each program targets a specific intelligence gap. ARCADE (Artificial Reasoning for Circuit Automation and Design Engineering) seeks to accelerate electrical circuit design by developing an AI-driven knowledge assistant that ingests technical data like schematics and datasheets. COSMIC (Commercial Observation for Spatio-temporal Monitoring for Indications of Change) aims to integrate commercial remote sensing data and open-source geolocation information into dynamic geospatial models. DECIPHER focuses on detecting and interpreting emerging or coded language, including slang, jargon, and acronyms, by generating contextual definitions and tracking how meaning evolves over time.

LocUS (Location Using Sound) develops capabilities to geolocate video content using audio and visual signals, improving analysts' ability to determine where footage was captured even when metadata is unavailable. MOVES (Movement Observation and Video-based Evaluation System) explores the use of video analytics to support remote assessment of neurological conditions, with potential applications in both healthcare and broader intelligence use cases.

The OTA-based approach matters for industry participation. IARPA's Emerging Technology Accelerator uses an other transaction authority model to reduce barriers to entry for commercial firms, enable iterative development and rapid prototyping, and support transition of technologies into operational environments. To qualify for an other transaction agreement, a prototype project must satisfy at least one of three criteria: significant non-traditional participation, small business and NDC exclusivity, or a cost-sharing requirement where at least one-third of total project cost is funded by non-federal sources.

Principal Deputy Director of National Intelligence Aaron Lukas stated these research programs will help build capabilities directly applicable to mission needs by bridging the technical gap between emerging solutions and successful application. Russell Miller, IARPA Director, explained the agency wants to be the front door for the intelligence community's emerging technology requirements to ensure they can harness private-sector expertise and accelerate breakthroughs in mission-critical technologies.

The effort aligns with the Office of the Director of National Intelligence's ODNI 2.0 initiative to accelerate the transition of emerging technologies into operational use and strengthen collaboration with the private sector. ExecutiveGov's coverage of the announcement provides additional context on the program structure and qualification requirements.

This represents a shift in how intelligence agencies engage with commercial innovation. The traditional procurement process often takes months or years to award contracts, with layers of bureaucracy that can stifle rapid iteration. The OTA model cuts through that (though it still requires navigating federal compliance requirements, which is its own form of obstacle course). Companies can now prototype faster, fail faster, and iterate without the usual red tape.

The physical reality of these programs matters for end users. An analyst working with LocUS won't just see a "geolocation feature"—they'll be clicking through video files, watching audio waveforms sync with visual cues, and waiting for the system to process audio fingerprints against known acoustic signatures. That processing time, the UI friction of uploading large video files, the texture of the interface itself—these details determine whether the tool actually gets used in the field or sits gathering digital dust.

Industry analysts note this positions IARPA differently from traditional defense procurement. The cost-sharing requirement means commercial firms must invest their own capital, which creates skin in the game but also raises barriers for smaller startups. The non-traditional participation clause opens doors for academic institutions and non-profits, but those entities often lack the infrastructure to scale prototypes into operational systems.

Whether these programs deliver on their promises remains to be seen. The intelligence community has a long history of ambitious research initiatives that never make it past the prototype phase. The real test isn't whether IARPA can fund these programs—it's whether the resulting technologies can survive the transition from research lab to operational deployment, where reliability and security requirements are orders of magnitude stricter than in academic settings.

The announcement itself is straightforward. The execution will be messy. Whether users actually pay for it—or in this case, whether intelligence agencies actually adopt these tools—remains the real question.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn