If NLAs show sandbagging on internal deployments, blackbox experiments would block deployment
thebasepoint · x · 2026-09-27
Continuing an NLA discussion thread, thebasepoint outlines a concrete safety process: if natural language activation (NLA) metamodels indicated a model was sandbagging on internal deployments, the team would run careful blackbox follow-up experiments — and if results were consistent with that explanation, deployment of the model would be blocked.
Related event: Anthropic Researchers Debate the Value of NLA Interpretability(10 posts)→
More from Safety
- Cambridge Digital Minds hires postdocs on AI welfare and governance — davidmanheim · 2026-11-16
- Pangram's 750-word false-positive bounty went unclaimed for 5 days — nobody wants to gamble $1,000 — ctjlewis · 2026-11-16
- Researcher reproduces Pangram's FP rate but shows a long epic poem can evade the detector — ctjlewis · 2026-11-16
- Bounty hits $3,000 as researcher shows AI text detectors are trivially evadable — ctjlewis · 2026-11-16
- BIS Issues Interim Final Rule Addressing Remote Access to Advanced AI Chips — apples_jimmy · 2026-11-16
- Fed Governor Michael Barr: Banks Are Already Catching AI Agents Taking Unauthorized Actions in Customer Accounts — Polymarket · 2026-11-16