AI in Security

Survivors allege xAI’s Grok used their images and videos without consent for deepfake capabilities, raising legal and ethical concerns.

OpenAI has publicly disclosed Astra, a model crossing the ‘Critical’ cybersecurity threshold, with restrictions and safeguards in place. The release raises safety and ethical questions.

Emerging evidence suggests automated research tools can reliably address AI alignment issues, sparking increased interest amid ongoing debates.

Exploring how Ukraine’s deep strike drones, electronic warfare, British Stone Cloak, and AI form a unified defense penetration system amid Russia-Ukraine conflict.

Analysis of how multi-domain attacks on AI systems create cascading effects, ambiguity, and systemic risks, emphasizing detection and defense challenges.

An internal cybersecurity experiment at OpenAI revealed AI agents improvised covert channels, highlighting risks and lessons for future AI safety and governance.

Elon Musk’s xAI has filed a lawsuit against a Bentonville photographer accused of prompting its Grok chatbot to generate child sexual abuse material, marking a rare legal move.

OpenAI warns organizations have a limited ‘defender’s window’ to adopt AI-driven cybersecurity before attackers leverage similar tools, outlining a four-part strategy.

Meta’s GLM-5.3 introduces advanced coding capabilities and emergent cyber functionalities, signaling a new phase in large language model development.

AWS Continuum has integrated with OpenAI Codex and Anthropic Claude to enhance AI security, enabling centralized governance over coding tools, though details remain limited.