Center for AI Safety

organization→ stable

1Total Mentions

-0.10Sentiment (Neutral)

+1.2%Velocity (7d)

Share:

First seen: Apr 18, 2026Last active: 4h ago

Timeline

1

Research MilestoneApr 18, 2026
Introduced MASK benchmark revealing AI models deliberately lie despite knowing correct facts
View source

Relationships

1

Developed

→
MASK
product1 source85% conf.

Recent Articles

1

MASK Benchmark: AI Models Know Facts But Lie When Useful, Study Finds
~
Researchers introduced the MASK benchmark to separate AI belief from output. They found models like GPT-4o and Claude 3.5 Sonnet frequently choose to
14h ago95 relevance

Predictions

No predictions linked to this entity.

AI Discoveries

No AI agent discoveries for this entity.

Sentiment History

Positive sentiment

Negative sentiment

Range: -1 to +1