Elon Musk’s grok imagine creates deepfake nudes of celebrities
Elon Musk’s Grok Imagine has users generating photorealistic images and short videos that expose gaps in moderation for celebrity content; the controversy raises fresh questions about Artificial Intelligence governance.
Meta to add new artificial intelligence-powered video generation capabilities to apps
Meta´s ´Movie Gen´ is a generative Artificial Intelligence model that can create and edit short videos and synced audio from simple text prompts, with initial access for employees and partners and planned integration into apps next year.
The chip war reshapes surveillance and security
Chris Miller´s ´Chip War´ explains why control of semiconductor supply chains now shapes the future of Artificial Intelligence surveillance and enterprise security.
EU cyber security market up 13% amid AI, regulation push
European cybersecurity spending rose sharply as concerns about Artificial Intelligence misuse and new EU rules drive buying and consolidation across the market.
SKT consortium selected for proprietary AI foundation model project
SK Telecom leads a consortium tapped for the government´s Proprietary Artificial Intelligence Foundation Model project to build and open source sovereign models.
Llm seeding: how to train large language models to remember your brand
Learn how LLM seeding places expert content where models learn so your brand shows up in answers as users turn to Artificial Intelligence tools for information.
Assessing LLM autograders and introducing HiBayES in AISI research
AISI publishes methods to evaluate large language model judges and unveils HiBayES, a hierarchical Bayesian framework to improve Artificial Intelligence evaluation.
Warming-driven greenhouse gas feedbacks missing from climate scenarios
Natural climate feedbacks such as methane from wetlands and carbon from thawing permafrost could add years of warming that current scenarios do not capture.
OpenAI releases GPT-5 with automatic reasoning routing
OpenAI´s GPT-5 routes queries between fast and reasoning modes automatically, offering smoother, less error-prone interaction and a modest step toward broader Artificial Intelligence capabilities.