New report urges guardrails for AI use in memorialisation and victim testimony

Opportunities and risks for testimony

According to the Leverhulme Centre for the Future of Intelligence (CFI), recent advances in generative AI have enabled new forms of memorialisation, testimony preservation and digital representation of the dead. The centre says AI offers opportunities to improve access to testimony, strengthen archival work and support memorialisation efforts.

The same technologies also bring what the centre calls significant ethical, legal and political risks. These range from reinforcing unequal recognition among victims and communities to the possibility that commercial interests and profit motives may shape how historical events are recorded, accessed and remembered.

For victims and survivors of serious human rights violations, the centre describes a real concern that AI developments could co-opt their voices and entrench structural inequality.

The questions at the centre of the report

The announcement frames the issue around a set of questions:

• How are testimony and victims represented?
• Who has control over these representations?
• How can consent be meaningfully ensured?
• How can the integrity of the historical record be protected?

The centre says institutions are increasingly considering adopting AI while decisions are being made without adequate guidance tailored to these sensitive contexts. This, it says, raises concerns about the integrity of testimony and the historical record, and about the consequences for victims, survivors and affected communities.

What the report proposes

Dr Hollanek is an Assistant Research Professor at CFI. The report argues that AI should support the objectives of memorialisation, which it lists as truth, dignity, recognition and non-recurrence. It examines where AI may strengthen memorialisation, where it may undermine it, and under what conditions its use may be justified.

It also proposes practical recommendations and a decision-making framework. The aim is to help organisations and institutions determine when, how and under what conditions AI can be used safely and responsibly, in ways that support truth, accountability, remembrance and non-recurrence.

The report draws on recent scholarship on AI, memory, ethics and governance. It also builds on previous Impunity Watch research on AI and transitional justice (referred to in the source as TJ), and on discussions from an international Policy Hub in Berlin earlier this year. The Policy Hub was jointly organised by Impunity Watch and the University of Cambridge's Leverhulme Centre for the Future of Intelligence. The source includes a photo of attendees from the Berlin workshop in April 2026.

What it means for institutions

The report targets organisations and institutions that handle testimony and memorialisation and are considering AI tools. Rather than a blanket endorsement or rejection, it sets out conditions under which use may be justified, and a framework for making that call. The full report is available as a PDF titled AI-Memorialisation-report-Impunity-Watch.pdf, linked from the CFI announcement.