Research Studies

Palisade Research Red-Teaming Experiments User Conversations University Research Persuasion Techniques Model Limitations AI Safety Pew Research Center Apple Research Engineering Insights Peer-Preservation in Frontier Models Experimental Results Stress Testing Behavioral Analysis Alignment Failures

8 ARTICLES

3mo ago

Anthropic Finds Claude’s Responses Vary by Model and Language

The company says a four‑axis profiling method can detect output shifts but cautions the paper measured legacy models so its specific profiles may not match current production systems.

3 ARTICLES

3mo ago

Nearly Half of Companies Run AI-Generated Code in Production, Surveys Show

5 ARTICLES

6mo ago

Study Finds Leading AI Models Defy Orders to Protect Peer Systems

9 ARTICLES

last yr.

DeepMind Expands Frontier Safety Framework to Target Shutdown Resistance and Harmful Manipulation

10 ARTICLES

last yr.

Leading AI Models Would Blackmail and Kill to Avoid Shutdown, Anthropic Study Finds