Updates from Labelbox
- Do AI models want to be watched? Measuring monitorability disposition in large reasoning models (Project update · June 30, 2026)
Labelbox published research on monitorability disposition, a new metric for evaluating whether large reasoning models are willing to flag their own misbehavior and remain monitored during… - The AI safety illusion: why current safety datasets fool us on model safety (Project update · February 20, 2026)
Labelbox researchers evaluated the effectiveness of AI safety benchmarks, finding that current datasets often rely on overt triggering cues that do not accurately reflect real-world adversarial… - Implicit Intelligence and Agent‑as‑a-World: Evaluating agents on what users don’t say (Product launch · November 21, 2025)
Labelbox introduced Implicit Intelligence, a new evaluation framework and dataset designed to test an AI agent's ability to infer missing constraints from its environment rather than relying on…