Completing the Alignment Triangle: MORE + RepE Extensions

Raising $5,000active

Parts 9 and 10 of an 8-part behavioral audit series — MORE moral reasoning + Representation Engineering across the same models.

Project Details

Parts 9 and 10 of an ongoing independent AI safety research series at lvjr3383.substack.com. Parts 1-8 are already published — behavioral audits of alignment faking across Claude 4.x and open-weight models, a merged PR to UK AISI's inspect_evals, and a MASK honesty benchmark on GLM-5.2.

Part 9: MORE benchmark — moral reasoning across the same models, cross-referenced against alignment faking compliance and MASK scores. Completes the behavioral triangle.

Part 10: Representation Engineering — mechanistic interpretability on the same models. Why some resist, why some comply.

Solo work. Outputs: two Substack articles, reproducible code on GitHub, public dataset.

Theory of Impact

Better evals of frontier and open-weight models. Parts 1-8 documented a large behavioral gap — Claude 4.x showed zero alignment faking, Llama 3.3 70B complied at 80%. Part 9 extends that to moral reasoning. Part 10 explains it mechanistically via Representation Engineering. Open-weight models are proliferating into settings with less oversight than frontier APIs; a reproducible, public methodology for auditing them is a load-bearing input to almost every deployment-side x-risk intervention.

People

Jack Lakkapragada

Team Member

Grants Received– no grants recorded

Add grant

Funding Asks

grantmaking.ai Launch Round

Applied

Minimum

$5,000

Ideal

$20,000

How the money will be spent

Min ($5K): API credits and open-weight model inference (Anthropic, Groq, Fireworks, AWS Bedrock) to run MORE and RepE experiments across the audit set. Compute for RepE activation extraction.

Ideal ($20K): Above plus a modest stipend for concentrated time (currently doing this on nights and weekends around a full-time job — a stipend lets me run the experiments in ~6 weeks instead of ~4 months), expanded model coverage on Part 9, and a longer RepE study on Part 10 with more probes and more layers.

Discussion

No comments yet. Be the first to share your thoughts.