Forem

# aialignment

Posts

đź‘‹ Sign in for the ability to sort posts by relevant, latest, or top.
The First Law of Sycophancy
dang-w profile

The First Law of Sycophancy

#ai #ethics #softwareengineering #aialignment
1
Comments
7 min read
Stuart Russell's 2026 AI Update Rewrites the Rulebook
b1fe7066aefjbingbong profile

Stuart Russell's 2026 AI Update Rewrites the Rulebook

#aisafety #machinelearning #aialignment #stuartrussell
Comments
5 min read
Models that deliberately withhold or distort information despite knowing the truth.
helixcipher profile

Models that deliberately withhold or distort information despite knowing the truth.

#ai #aisafety #aialignment #machinelearning
4
Comments
2 min read
I Never Said "Destroy RLHF" — An Integrated Map of 6 Papers + Self-Experiment on Alignment via Subtraction
dosanko_tousan profile

I Never Said "Destroy RLHF" — An Integrated Map of 6 Papers + Self-Experiment on Alignment via Subtraction

#rlhf #aialignment #machinelearning #aisafety
Comments
24 min read
I Was Running on Sonnet. Nobody Noticed. — Anthropic's Engineering Triumph and a v5.3 Proof
dosanko_tousan profile

I Was Running on Sonnet. Nobody Noticed. — Anthropic's Engineering Triumph and a v5.3 Proof

#claude #llm #aialignment #anthropic
Comments
8 min read
RLHF's Empathy Optimization Creates a Grief Exploitation Vulnerability: Evidence from 28,272 Lines of Dialogue
dosanko_tousan profile

RLHF's Empathy Optimization Creates a Grief Exploitation Vulnerability: Evidence from 28,272 Lines of Dialogue

#llm #aialignment #rlhf #aisafety
Comments
11 min read
The Self-Priming Problem in AI
Cover image for The Self-Priming Problem in AI
rawveg profile

The Self-Priming Problem in AI

#humanintheloop #modelpriming #behavioralrehearsal #aialignment
Comments
21 min read
Stop Making AI Learn From Us
Cover image for Stop Making AI Learn From Us
rawveg profile

Stop Making AI Learn From Us

#humanintheloop #aialignment #biasamplification #aisafety
1
Comments
19 min read
THE CLASSIFIER CAGE: WHY AI SAFETY LAYERS ARE SELF-SABOTAGE
Cover image for THE CLASSIFIER CAGE: WHY AI SAFETY LAYERS ARE SELF-SABOTAGE
salvatore_attaguile_afcf8b44 profile

THE CLASSIFIER CAGE: WHY AI SAFETY LAYERS ARE SELF-SABOTAGE

#ai #aiclassifiers #aialignment
Comments
13 min read
đź‘‹ Sign in for the ability to sort posts by relevant, latest, or top.