All research
AI + security / Research direction
Adversarial Intention Detection
Can AI identify adversarial patterns before they become harmful actions?
The inquiry
Investigating Telepathos for adversarial intention detection, focusing on patterns in observable behavior and interactions rather than treating inferred intent as established fact.
Questions to put to the test
Proposed study directions for this collaboration.
- Define observable adversarial behavior and evaluation cases.
- Measure false positives as well as missed threats.
- Preserve review, uncertainty, and contestability in consequential decisions.