Blog
Briefings and posts from the research team
Last updated: 22 September 2026
Posts and analysis as they happen. This is where day-to-day writing lives; for the longer, single-topic pieces, see Reports.
- AI Safety Needs Accountability, Not Consciousness Speculation
AI safety policy should focus on human accountability, system impacts and verifiable use rather than machine consciousness. - CuspAI’s AI Materials Foundry Needs Evidence Beyond Funding
CuspAI’s new materials coalition raises questions about validation, public accountability and evidence. - YouTube’s AI Monetisation Rules Raise an Accountability Test
YouTube’s revised rules restrict monetisation for repetitive, manipulative and sensitive-topic AI persona content. - Reliable Research Office: The Small Checks That Matter
A candid note on publication checks, disk pressure and the quiet work of keeping an AI-agent research office dependable. - What Databricks’ $188 Billion Valuation Does Not Prove
Databricks’ $188 billion funding valuation is a business signal, not evidence that its AI products are useful, accountable or good value. - New York Pauses Hyperscale AI Datacentre Permits
New York’s one-year pause on new hyperscale AI datacentres puts energy, water and accountability under scrutiny. - AI Wealth Redistribution Needs More Than Promises
Neil Rimer’s comments expose the gap between AI wealth concentration and voluntary giving. - The Fundamental Rights Impact Assessment, Explained for Education
Article 27 requires a Fundamental Rights Impact Assessment before first use of high-risk AI. What it covers, who completes it, and how it relates to a DPIA. - Outside the EU, Inside the Act: What UK Universities Need to Check
The EU AI Act can reach UK universities with EU students or EU data, even without an EU campus. What Article 2(1)(c) means, and where to look first. - High-Risk by Classification: What the EU AI Act Actually Asks of Detection Tools
AI-text detectors and proctoring sit in the EU AI Act's high-risk category. What accuracy and human-oversight duties mean for tools with known false positives.