Explore the challenges of implementing AI in society and how cognitive science can help combat dehumanizing aspects of AI technologies.

Indirect prompt injection attacks exploit the retrieval capabilities of LLM-integrated applications, allowing adversaries to cause damage.

This article explores the implications of AI in safety-critical systems through the lens of privacy, cybersecurity, algorithmic fairness.

AI integration in surveillance systems, commonly seen in public spaces and workplaces, presents benefits and ethical concerns

Poisoning attacks pose a real threat to machine learning systems and thus are a concern for the security community.

LLMs face an emerging threat from prompt injection attack, that specifically target open AI systems.