Applies to: All QPF and Quantum Policy Radar operations · Review cadence: Semi-annual
Position
QPF uses artificial intelligence tools as part of its research, drafting, and analytical workflow. We believe disclosure about how we use these tools is more credible than pretending they do not exist. This policy explains what AI does and does not do in QPF’s work.
How QPF uses AI
AI tools assist QPF’s editorial team in several ways. We use them to accelerate initial research by synthesising large volumes of source material across jurisdictions. We use them to assist with early-stage drafting, where an AI-generated first pass is then substantially rewritten, restructured, and fact-checked by our editorial team. We use them to check internal consistency across our analytical outputs and to identify gaps in coverage.
What AI does not do
AI does not make editorial judgements for QPF. The assessment of what a development means, how significant it is, what remains uncertain, and what subscribers should pay attention to is the work of human analysts with domain expertise. No QPF or Quantum Policy Radar content is published without substantive human editorial review, rewriting, and approval.
AI does not determine what QPF covers or how topics are framed. Coverage decisions, editorial priorities, and analytical framing are made by QPF’s editorial team based on our coverage mandate, our sourcing standards, and our understanding of what matters to our readers.
AI does not have access to subscriber data, confidential event discussions, or proprietary information shared with QPF under embargo or confidence.
Labelling
QPF does not label individual pieces of content as “AI-assisted” because AI assistance is embedded throughout our workflow in the same way that search engines, databases, and analytical software are embedded in any modern research operation. The distinction between “AI-assisted” and “not AI-assisted” is not meaningful when AI is used as a research tool rather than as an autonomous author.
If QPF’s use of AI changes materially, or if new AI capabilities alter the nature of our workflow in ways that affect the analytical product, this policy will be updated to reflect those changes.
Data and confidentiality
QPF evaluates the data handling practices of any AI service it uses. We do not input subscriber data, confidential event content, embargoed source material, or personally identifiable information into third-party AI services unless those services provide contractual guarantees against data retention and training use. We review these guarantees periodically.
QPF’s internal AI use guidelines specify which tools are approved for which tasks, what categories of information may and may not be processed through external AI services, and what disclosure obligations apply. These internal guidelines are reviewed semi-annually.
Our view
AI is a tool. Used well, it allows a small editorial team to cover more ground, process more sources, and deliver analysis faster. Used poorly, it produces generic, uncritical output that experienced readers recognise immediately. QPF’s editorial standards exist precisely to ensure we use AI well: as an accelerant to human expertise, not as a substitute for it.
We welcome questions about our use of AI. Readers who wish to understand our workflow in more detail are encouraged to contact us at [email protected].