CCTest · Blog
Concept Unlearning in Text-to-Image Models Needs Certification, Not Just Attack Tests
AI Safety
cctest.ai
AI Safety

Concept Unlearning in Text-to-Image Models Needs Certification, Not Just Attack Tests

A new study argues that automated adversarial prompt searches can substantially underestimate residual concept leakage in text-to-image diffusion models. Its certification framework provides confidence-aware upper bounds instead of relying only on observed attack success rates.

Read more
CCTest · Blog
AI Is Accelerating Mathematics—But Mathematicians Fear It May Change More Than the Speed of Solving
AI for Science
cctest.ai
AI for Science

AI Is Accelerating Mathematics—But Mathematicians Fear It May Change More Than the Speed of Solving

According to the source material, 25 Fields Medalists have endorsed a statement warning that some AI companies are reducing mathematical research to benchmark performance. Their concern is not the use of AI itself, but the preservation of understanding, attribution, education, and open collaboration.

Read more