FACT Aotearoa News Briefs 6 July 2026

AI-generated impersonations of political figures are judged by members of the public to be more authentic, relevant and coherent than the speakers' actual debate responses - University of Passau
In a new study, researchers used GPT-4 Turbo to generate impersonated responses to audience questions drawn from 30 episodes of BBC1's Question Time. The study conclusively shows that "... humans think AI-generated debate content is more authentic than what the actual well-known public people said."
https://phys.org/news/2026-06-ai-generated-debate-outscore-real.html

Safety Planning Toolkit - A practical guide for doing great work even when things feel scary - Tāhoho Trust, Amnesty International, and Anchordown
This toolkit unpacks the growing issue of online harm aimed at individuals and organisations. It’s designed to provide information, ideas and resources to help you feel empowered to do important work — even in a challenging environment.
https://safetytoolkit.nz

MIT study shows reliance on AI diminishes human ability to distinguish real from fake news
After a month of using AI to help people distinguish fake news from real news, their ability to perform the task unaided by AI diminished by an average of 15%.
https://www.fastcompany.com/91558827/ai-impact-on-brain-cognitive-ability-mit-study-reveals-more-troubling-data

It's possible to predict when a conspiracy theory will jump from one platform to another - USC
"Discourse networks" of users engaged in the same kind of content can predict the path of disinformation flows with 94% accuracy.
https://www.isi.edu/news/85501/what-if-we-could-see-disinformation-coming-usc-scientists-say-we-can/

A Policy Playbook to Inoculate the Public Against AI Text Falsehoods - Tech Policy Press
People are getting better at recognising AI-generated false images, but worse at recognising AI-generated text. The solution: Give people the knowledge and context they need before they encounter misleading content, not after - prebunking.
https://www.techpolicy.press/a-policy-playbook-to-inoculate-the-public-against-ai-text-falsehoods/

Young men in the UK who regularly play video games are twice as likely as the general population to believe conspiracy theories - University of Arts London
This distrustful, disengaged group gets much of its news through YouTube, podcasts and social media rather than mainstream outlets, and plays games more than any other segment of society to escape daily hardships. The report suggests it's the wider ecosystem surrounding gaming—not the games themselves—that leaves them vulnerable.
https://www.videogamesindustrymemo.com/p/young-male-gamers-are-twice-as-likely

The Montréal shooter’s manifesto isn’t left or right — it’s rooted in misogyny - The Conversation
A lone attacker builds a world view the way you build a plate at a buffet — a bit of racism here, some antisemitism there, a scoop of anti-government conspiracy; whatever fits his grievance. It doesn’t hold together as a doctrine. It holds together as a feeling.
https://theconversation.com/the-montreal-shooters-manifesto-isnt-left-or-right-its-rooted-in-misogyny-286192