IARPA Explores AI to Reimagine Physical Security and Combat Misinformation

HAYSTAC seeks to proactively monitor and warn citizens

One AI-powered project that Marsh teased is the aptly named HAYSTAC, or Hidden Activity Signal and Trajectory Anomaly Characterization.

It aims to develop “capabilities that produce large-scale microsimulations so you can understand human motion and build AI reasoning engines that can identify anomalous motion trajectories and generate normal trajectories,” Marsh told the symposium audience. Put more bluntly, given the explosion of data gathered from Internet of Things devices and smart city technology, it is possible to better understand normal and abnormal patterns of human movement. With the help of AI, the government could develop more efficient and proactive public safety systems and processes.

“As we move forward, how can you detect these things happening so that we can figure it out in advance and create an early warning?” Marsh said.

This could be useful during a natural disaster or terrorist attack, for example, when most people are fleeing areas as first responders move in.

EXPLORE: New technologies at the service of agencies and citizens.

“Disturbances to roads, bridges or critical infrastructure elements create traffic and force people to reroute,” according to an IARPA press release. “An increase in civil unrest changes the daily way of life, as health and basic needs take priority over work and leisure. Often the disruptive events themselves are not directly observable, but an understanding of surrounding motion anomalies can inform rapid response and analysis.

HAYSTAC is in the general agency announcement period through July 1, during which IARPA invites research proposals from private sector companies. The program is expected to last four years, starting in October and ending in March 2026.

“While achieving HAYSTAC will be a multi-year process, once it is complete we will have reframed the way we view business around the world,” HAYSTAC program manager Jack Cooper said in the press release. . “And it won’t be a static concept of where things are on a map, but a dynamic concept based on how they move and what’s out of the ordinary.”

HIATUS helps identify and combat misinformation campaigns

As AI is put to work improving physical safety and security, it is also exploring how to better sort through and understand intangible things, like human expression. IARPA’s Human Interpretable Attribution of Text Using Underlying Structure (HIATUS) program aims to “develop novel human-usable AI systems for attributing authorship and protecting the privacy of authors through the identification and exploiting explainable linguistics,” helping to solve a problem for intelligence agencies, she said.

Timothy McKinnon, the IARPA program manager spearheading this work, recently told Nextgov, “Just think if you had 100 different people, and you asked them to describe something simple — like how to open a door — by two sentences or one sentence. You would probably get around 100 different answers, right? And each person somehow has their own idiosyncrasies as an author that are potentially used by systems of attribution of authorship.

LEARN MORE: Swarm intelligence: what is it and how are agencies using it?

The program aims to identify stylistic features, such as syntax, that can help identify who wrote a given text, much like a fingerprint, McKinnon said.

“The technology would be able to identify that fingerprint against a corpus of other documents and match them if they come from the same author,” he said. “On the privacy side, the technology would find ways to alter the text, so that it no longer resembles a person’s handwriting.”

The HIATUS effort has the potential to be a game-changer to track misinformation campaigns and help combat human trafficking and other malicious activity in online text forums, McKinnon told the outlet.