Darpa Unveils New Gard Program to Combat Adversaries Who Try to Trick’ Advanced U.s. Tech
DARPA has pulled back the curtain on the GARD program, which counteracts adversary efforts to trick America's AI defense systems.
The Debrief 7:22 pm on April 10, 2024
The Defense Advanced Research Projects Agency (DARPA) has unveiled its new Guaranteeing AI Robustness against Deception (GARD) program, aimed at enhancing AI systems to counteract adversarial attacks. Adversaries can deceive advanced defense technologies by using low-tech methods to trick AI targeting systems into misclassifying friend or foe vehicles. GARD is focused on building defenses against such attacks and has partnered with industry leaders to develop tools to anticipate and counteract these vulnerabilities. The capabilities could potentially be transitioned across the entire defense department ecosystem, as well as to non-military partners in the future.
- DARPA introduces GARD program to enhance AI systems against adversarial attacks
- Low-tech methods used to trick AI targeting systems can be a threat to defense technologies
- GARD focuses on building defenses and partners with industry leaders for tools development
- Capabilities could potentially be transitioned across defense department ecosystem and to non-military partners
- DARPA's core mission is to prevent and create strategic surprises, implying ongoing research in AI advancements
https://thedebrief.org/darpa-unveils-new-gard-program-to-combat-adversaries-who-try-to-trick-advanced-u-s-tech/