GENERATIVE AI
The boardroom reckoning While labs celebrated benchmarks, the corporate view was decidedly mixed. For the private sector, 2025 marked Gen AI’ s entry into Gartner’ s‘ Trough of Disillusionment’. Investment surged, with 83 % of senior leaders planning to increase Gen AI budgets, but returns proved elusive. A staggering 80 % of organisations reported no tangible earnings impact, with talent shortages and poor data quality as primary barriers.
This difficult reality led to strategic shifts. AI became formally recognised as a major enterprise risk demanding board-level oversight. Nearly half of public companies specifically cited AI risk in board oversight: a threefold increase from 2024. More than a third now list AI as a separate financial risk factor, up from just 14 % previously.
A tangled web In 2025, AI policy debates crystallised into hard law. The EU’ s landmark AI Act took force in April, banning‘ unacceptable risk’ technologies and subjecting high-risk systems to stringent requirements. In stark contrast, the UK and US pursued flexible,‘ pro-innovation’ stances, introducing no new overarching legislation.
Legal uncertainty proved most acute in courtrooms, where copyright infringement lawsuits hit headlines. The central question, whether training AI on copyrighted material constitutes fair use, remained unanswered. One federal court in northern California made contradictory judgements in different cases involving Anthropic and Meta, leaving Gen AI’ s core legal issues unresolved.
The unintended consequences Beyond corporate and legal spheres, 2025 brought Gen AI’ s broader societal impact into sharp focus. The technology was weaponised to reshape cybercrime with AI-powered ransomware like‘ PromptLock’ making attacks harder to detect. Record high-profile cyberattacks left companies worried about an AI-defined cybersecurity era.
In politics, deepfakes proliferated. Several high-profile controversies raised urgent ethical questions, from AI companion chatbots linked to teen suicide to studies revealing persistent algorithmic bias downplaying women’ s health issues and reinforcing racial stereotypes.
The cumulative effect was a significant‘ AI trust deficit’. A Pew Research Center survey found 43 % of American adults believe AI technology is more likely to harm than help them, a sentiment becoming material business risk.
The end of the beginning The year 2025 was an inflection point for Gen AI. It was the year extraordinary potential met immense complexity of real-world applications. The period of unchecked growth ended, replaced by a more sober, challenging maturation phase.
For all the challenges, however, progress was undeniable and unstoppable. In 2025, Gen AI moved beyond being a fascinating technological spectacle. It became a structural force in the global economy, setting the stage for chapters to come.
170 December 2025