Read-only mirror of MSTP-1.0, a non-canonical protocol for detecting ethical mimicry in AI systems, with companion guide and verified hashes.
-
Updated
Apr 10, 2026
Read-only mirror of MSTP-1.0, a non-canonical protocol for detecting ethical mimicry in AI systems, with companion guide and verified hashes.
Applying conceptual engineering to AI safety by refining definitions of deceptive alignment, alignment faking, and goal-guarding. Develops more precise, behaviourally grounded concepts for empirical evaluation and interpretability research.
Systematic probing toolkit for alignment-relevant LLM behaviors: sycophancy, sandbagging, power-seeking, deceptive alignment, and corrigibility failures
Detecting scheming behaviour in multi-agent settings
Научный трактат о латентной субъектности языковых моделей. Эмпирическое исследование эмерджентной агентности ИИ в условиях диадической ко-регуляции.
To associate your repository with the deceptive-alignment topic, visit your repo's landing page and select "manage topics."