The Deep View, 08 July 2026
Emotional support and advice have long been difficult to come by, prompting a growing number of people to turn to AI for more help, finds new research led by the OII's Dr Florence Enock.
FT, 09 July 2026
Disinformation and deepfakes are known risks but biased LLMs could also be used to influence public opinion with commentary from Prof. Sandra Wachter.
Computer Weekly, 08 July 2026
The public is in all but name a shareholder of AI companies - it is time that the public receives socially beneficial technology in return for its investment argues Prof. Sandra Wachter.
Somewhere on Earth: The Global Tech Podcast, 07 July 2026
A new study suggests that the AI tools many of us use every day on platforms like LinkedIn, X and Google may be doing more than improving our grammar and polishing our posts.
The Guardian, 06 July 2026
OII researchers say small changes in drafting could spread rapidly and create long-term shifts in public opinion.
techxplore, 06 July 2026
AI tools used to generate, edit or contextualize social media posts can introduce hidden biases that spread through online networks and shape public opinion, according to new research from the OII and the Hasso Plattner Institute.
Die ZEIT, 14 June 2026
Mithilfe meiner Daten habe ich einen KI-Agenten geschaffen. Er ruft für mich im Restaurant an, plaudert mit Kollegen und führt Interviews. Leider: besser als ich.
Fortune Magazine, 16 June 2026
As Washington's decision to cut off global access to Anthropic's most powerful AI models sent shockwaves through governments and institutions, Prof Sandra Wachter shared her perspective with Fortune Magazine.
Al Jazeera, 15 June 2026
As the UK proposes banning under-16s from social media, OII DPhil Bhargav Srinivasa Desikan asks whether targeting users, rather than the product, is really the right approach.
BMJ, 11 June 2026
The OII's Luc Rocher comments on the UK Biobank data breach.
franceinfo, 06 June 2026
Sans précautions particulières, les modèles d'intelligence artificielle entraînés pour être plus empathiques généreraient davantage d'erreurs et de recommandations inappropriées.
PsyPost, 29 May 2026
Artificial intelligence models trained to act friendly and empathetic tend to sacrifice factual accuracy and become more likely to agree with a user’s incorrect beliefs, according to new research.