Q: What was VECTOR and why did it attract attention? A: VECTOR was an unofficial AI application built by a noncommissioned officer on Army Vantage that promised to help soldiers draft OERs and NCOERs and prepare for promotion boards. The Army suspended it while conducting an Army AI compliance review after promotional messages claimed it could tap historic board data and it spread quickly across units. Q: Who created VECTOR and where was it hosted? A: The article says an individual noncommissioned officer created VECTOR and hosted it on Army Vantage, a Palantir-made platform that meshes data repositories with machine learning. The tool was not officially sanctioned by the Army and was later suspended for review. Q: What does the Army AI compliance review look at? A: An Army AI compliance review probes how the service manages shadow tools, what data flows are allowed on platforms like Army Vantage, and whether security, data access and policy requirements are met. It aims to balance rapid experimentation with oversight so unauthorized or high-consequence tools do not slip into use. Q: Did VECTOR actually have access to historical promotion board data? A: Army spokesperson Cynthia Smith said VECTOR did not have access to historical or sensitive data, contradicting the promotional message’s claims. That discrepancy over access and capability was a core risk signal that prompted the Army AI compliance review. Q: What potential benefits did experts say a tool like VECTOR could offer? A: Experts noted that a well-functioning tool could save time and help less-skilled raters produce clearer, more consistent evaluation language, which might reduce bias in assessments. They also warned such models require validation and verification testing to ensure accuracy and secure behavior before being used operationally. Q: What are the main risks the Army highlighted in cases like VECTOR? A: The main risks include unauthorized access to personally identifiable information or historical records, potential security vulnerabilities or backdoors in networks, inaccurate or biased outputs, and the spread of shadow IT that can undermine trust in promotion decisions. An Army AI compliance review is intended to uncover and mitigate these kinds of risks before tools influence talent management. Q: What practical guardrails did the article recommend for responsible AI adoption? A: The article recommends measures such as standing up an approved sandbox inside Vantage, a light approval flow (idea → data check → security scan → pilot), and a catalog of approved datasets with immutable logs for auditability. It also advises model validation and red‑teaming, publishing a short model card, user training, mandatory disclosure when AI drafts text, and keeping humans in the loop for leader development. Q: If a soldier or leader encounters an AI tool like VECTOR, what actions should they take now? A: Soldiers and leaders should verify that a tool is authorized before use, avoid uploading PII, and use only approved datasets while keeping a human in the loop for evaluation language and board preparation. They should also document inputs and outputs, report promising tools to their chain of command for review or piloting, and coach subordinates on how AI suggestions align with actual performance.