Project Maven
Technology: ce
| ← Previous revision | Revision as of 00:56, 20 April 2026 | ||
| Line 70: | Line 70: | ||
As of September 2025, the director of the NGA claimed that by June 2026, Maven will begin to transmit "100 percent machine-generated" intelligence to combatant commanders using [[Large language model|LLM]] technology. [[Booz Allen Hamilton|Booz Allen]] was the defense contractor responsible for the LLM-integration phase.{{Cite web |date=2025-09-02 |title=The AI Doomsday Machine Is Closer to Reality Than You Think |url=https://www.politico.com/news/magazine/2025/09/02/pentagon-ai-nuclear-war-00496884 |access-date=2025-10-02 |website=POLITICO |language=en}} |
As of September 2025, the director of the NGA claimed that by June 2026, Maven will begin to transmit "100 percent machine-generated" intelligence to combatant commanders using [[Large language model|LLM]] technology. [[Booz Allen Hamilton|Booz Allen]] was the defense contractor responsible for the LLM-integration phase.{{Cite web |date=2025-09-02 |title=The AI Doomsday Machine Is Closer to Reality Than You Think |url=https://www.politico.com/news/magazine/2025/09/02/pentagon-ai-nuclear-war-00496884 |access-date=2025-10-02 |website=POLITICO |language=en}} |
||
In 2025 November, in a disaster response exercise by the [[Alaska Army National Guard]], Maven used publicly available datasets such as road systems and [[USGS Earthquake Notification Service| |
In 2025 November, in a disaster response exercise by the [[Alaska Army National Guard]], Maven used publicly available datasets such as road systems and [[USGS Earthquake Notification Service|US Geological Survey earthquake data]] layers into its [[geospatial intelligence]] to produce the common operating picture{{what}}. It was also integrated with [[Incident Command System]] Form 213RR, which allowed teams to track requests for assistance and match them with [[Title 10 of the United States Code|Title 10]] or [[Title 32 of the United States Code|Title 32]] resources.{{cite web |last=LaCount |first=Seth |date=2025-07-11 |title=AKNG COP manager leads charge on AI integration in disaster response |url=https://www.dvidshub.net/news/542602/akng-cop-manager-leads-charge-ai-integration-disaster-response |access-date=2026-03-06 |website=Defense Visual Information Distribution Service |publisher=U.S. Department of Defense |location=Joint Base Elmendorf–Richardson, Alaska, United States}} |
||
Cameron Stanley, the Pentagon's chief digital and artificial intelligence officer, said at a 2026 conference AIPCon 9 that Maven was being "deploying across the entire department".{{Cite AV media |url=https://www.youtube.com/watch?v=yrtDgoqWmgM |
Cameron Stanley, the Pentagon's chief digital and artificial intelligence officer, said at a 2026 conference AIPCon 9 that Maven was being "deploying across the entire department".{{Cite AV media |url=https://www.youtube.com/watch?v=yrtDgoqWmgM |title=Multi-Domain AI: The Future of Command and Control {{!}} CDAO at AIPCon 9 |date=2026-03-13 |last=Palantir |access-date=2026-03-15 |via=YouTube}} Maven has an AI Asset Tasking Recommender that can propose which bombers and munitions should be assigned to which targets. Maven is integrated with Palantir's Artificial Intelligence Platform (AIP), and Palantir's Army Intelligence Data Platform, an intelligence platform sold to the US Army. |
||
During the [[2026 Iran war]], Maven allowed the striking of over 1000 targets in the first day. It was planned that future Maven would allow striking over 1000 targets in an hour.{{cite news |last=Manson |first=Katrina |date=2026-03-12 |title=‘God, It’s Terrifying’: How the Pentagon Got Hooked on AI War Machines |url=https://www.bloomberg.com/news/features/2026-03-12/iran-war-tests-project-maven-us-ai-war-strategy |access-date=2026-03-06 |newspaper=Bloomberg Businessweek}} Maven's computer vision system increased the rate of target-per-day from |
During the [[2026 Iran war]], Maven allowed the striking of [[List of attacks during the 2026 Iran war|over 1000 targets in the first day]]. It was planned that future Maven would allow striking over 1000 targets in an hour.{{cite news |last=Manson |first=Katrina |date=2026-03-12 |title=‘God, It’s Terrifying’: How the Pentagon Got Hooked on AI War Machines |url=https://www.bloomberg.com/news/features/2026-03-12/iran-war-tests-project-maven-us-ai-war-strategy |access-date=2026-03-06 |newspaper=Bloomberg Businessweek}} Maven's computer vision system increased the rate of target-per-day from less than 100 to 1000. After integrating LLMs, the rate increased to 5000 targets per day. |
||
== Contractors == |
== Contractors == |
||
