“This work represents a major step ahead in strengthening our data benefit as we fight refined disinformation campaigns and synthetic-media threats,” says Bustamante. Hive was chosen out of a pool of 36 firms to check its deepfake detection and attribution expertise with the DOD. The contract might allow the division to detect and counter AI deception at scale.
Defending towards deepfakes is “existential,” says Kevin Guo, Hive AI’s CEO. “That is the evolution of cyberwarfare.”
Hive’s expertise has been skilled on a considerable amount of content material, some AI-generated and a few not. It picks up on indicators and patterns in AI-generated content material which can be invisible to the human eye however might be detected by an AI mannequin.
“Seems that each picture generated by one in every of these mills has that type of sample in there if the place to search for it,” says Guo. The Hive staff continually retains observe of latest fashions and updates its expertise accordingly.
The instruments and methodologies developed via this initiative have the potential to be tailored for broader use, not solely addressing defense-specific challenges but additionally safeguarding civilian establishments towards disinformation, fraud, and deception, the DOD mentioned in an announcement.
Hive’s expertise offers state-of-the-art efficiency in detecting AI-generated content material, says Siwei Lyu, a professor of laptop science and engineering on the College at Buffalo. He was not concerned in Hive’s work however has examined its detection instruments.
Ben Zhao, a professor on the College of Chicago, who has additionally independently evaluated Hive AI’s deepfake expertise, agrees however factors out that it’s removed from foolproof.
“Hive is definitely higher than many of the industrial entities and a number of the analysis methods that we tried, however we additionally confirmed that it’s not in any respect laborious to avoid,” Zhao says. The staff discovered that adversaries might tamper with photos in a method that bypassed Hive’s detection.