Artificial intelligence is no longer a futuristic concept on university campuses. For the overwhelming majority of students, it has become an integral part of everyday academic life—whether for research, learning, writing or programming. The latest CHE DataCHECK, Artificial Intelligence in Higher Education, reveals just how deeply AI has become embedded in students' daily routines. At the same time, the report exposes a widening gap between how students actually use AI and the support universities provide.

Until recently, many organisations followed a simple rule: if you wanted the best possible results, you used the most powerful AI model available. The more capable the model, the better the output—regardless of the cost. That mindset is now beginning to change. A new term is gaining traction across the AI industry: Modelmaxxing.

With ChatGPT Work, OpenAI is pursuing one of the most significant strategic shifts in the history of ChatGPT. Rather than simply answering questions or generating text, the new capability is designed to take ownership of entire workflows. ChatGPT Work is conceived as an agentic work assistant that can plan projects independently, combine information from multiple sources, interact with applications and ultimately deliver finished outcomes.

The music industry is responding to the growing flood of synthetic songs with a new labelling system. In future, two visual labels will indicate whether a sound recording was created primarily by generative AI or simply produced with AI assistance. "AI-generated" and "AI-assisted" are intended to give listeners an immediate understanding of how extensively artificial intelligence contributed to what they hear.

Germany’s dubbing industry is facing a conflict that goes far beyond the question of whether an artificial voice sounds natural enough. At its core, the debate is about the future of an entire profession, the right to one’s own voice and whether streaming platforms should replace cultural quality with automated mass production.

Following Prompt Engineering, Context Engineering and Harness Engineering, another concept is rapidly gaining traction in the AI world: Loop Engineering. It is becoming one of the most talked-about ideas within the communities building AI agents and next-generation coding assistants. The underlying principle is simple. Instead of guiding an AI agent step by step through a task, you design a system that enables it to work independently, evaluate its own progress and continue iterating until a clearly defined objective has been achieved.

The legal battle between Midjourney and the Hollywood studios Disney, Universal and Warner Bros. Discovery has evolved into far more than a conventional copyright dispute. While the studios accuse the AI company of using copyrighted characters and creative works to train its image generation models without permission, Midjourney has launched an unexpected counterstrategy: it wants the studios to reveal how they use artificial intelligence themselves.

Microsoft is reportedly evaluating whether DeepSeek could become part of Copilot Cowork. At first glance, this may look like a technical model-selection decision. In reality, it reflects something much bigger: the economics of AI agents, Microsoft's dependence on OpenAI and Anthropic, growing pressure from lower-cost Chinese models, and the question of how much geopolitical risk businesses are willing to accept in exchange for cheaper AI.