Dancing with the Devil

The question that arises from delivering the Truth Matters Project is, does AI have a place in the information stack? I wanted to offer some clarification as to what’s guided my decision.

The AI conundrum and public sentiment against the technology have played a major role in this project's decision-making. The serious public distrust surrounding this technology is well-founded, not just from a technology perspective, but also because of the threat it poses to jobs and the environment. America could not have had a worse leader in office to deal with this potential existential threat.

AI is not new; it’s been around and part of our lives in the background for a long time. It just wasn’t visible, and it hadn’t been made available for the average person to utilise. ChatGPT changed all that, and that’s when the experiment escaped the lab. The genie was released from the bottle, and there’s no putting it back.

So we have choices to make, and those choices should shape the regulatory framework. The problem is, Trump sees only one thing: a chance to stuff his pockets before the bubble bursts. He’s bet the house on AI, and he can’t afford to lose. The AI war with China is a whole other post, but the bottom line is that this administration has no appetite to do what’s required, and that’s a BIG problem.

I want to leave you in no doubt that I take the responsibility that comes along with The Truth Matters Project VERY seriously. This is not driven by a quest for wealth; indeed, when I can no longer administer it, I commit here and now to giving it to the next custodian for FREE. This is not about me; it’s not about seeking validation; it’s about creating a resource built on integrity and moral decency, grounded in fairness and the belief that nobody should be left behind and no child should go to bed hungry. Lofty ambitions, I know, but it is what it is, and I will guard its integrity with every fibre of my being. Trust is the currency of this project; without it, the project is worthless.

I feel so passionate about it that I made an important decision today: to separate my political position from the project's core ethics and values engine. My writing and opinions will be part of the site, but they'll live in the Reading Room, not the Library, which is reserved for primary source information. My passion for social justice can’t be separated from who I am, so I'll be hosting my opinions at martinmcsweeney.org to distinguish opinion from primary source data.

Why is this important? Because I want to play my part as an advocate for social justice without compromising the project's core ethical values, and to keep the site a home for uncontaminated FACTS. My body of work will be available in the site's Reading Room, alongside links to organisations such as MeidasTouch Network and Democracy Docket and other voices that align with my political and social justice values.

So why not base my advocacy work in the Reading Room? Because part of my wider vision is to provide a space to promote advocacy organisations and give them digital meeting spaces and access to integrated webinars. Doing that within the Truth Matters site would breach the project's ethical framework. For this reason, martinmcsweeney.org will be an independent site offering lots of exciting free tools for nonprofits and advocacy groups that align with our joint values. It will be as sophisticated and high-production-value as the Truth Matters site; it will just be a separate, independent resource.

Moving on, AI is not conscious, and since I’m not inside the labs creating it, I have no idea how close they are to achieving this goal. I want to make you aware that I deliberately chose Anthropic’s Claude to partner on the primary AI-assisted framework development that underpins the project. I spent the best part of three days deciding which model to use.

I want to leave you with no doubt surrounding the seriousness with which I am approaching this. I’m in a unique position to understand messaging and global and domestic politics on both sides of the Atlantic. I also possess a depth of technical knowledge that enables me to bring these elements together under a single roof- my own. No negotiating with competing interests or hidden agendas, just me, my values and the trust I have earned from writing over 500 in-depth pieces on Substack.

Please take a moment to watch the video below. I quote from the YouTube summary:

Chloe Lubinski leads Anthropic's research partnerships with the world's faith and philosophy traditions, convening scholars, clergy, philosophers, and ethicists alongside the company's most senior researchers to help inform the moral formation of AI. She trained in cognitive neuroscience at UC Berkeley, completed four years of Ignatian spiritual direction formation, and holds a graduate degree at the intersection of theology and psychology from Trinity College Dublin.

This is the model I have chosen to work with, and here's a link to the slide deck for the presentation, in case you want to view it as she speaks.

I am also including the video below from the July 2026 EDP 50th Anniversary Gala Dinner, featuring Tom Parker, journalist and host of The Next Five, the Financial Times’ business podcast, and Professor Yuval Noah Harari: historian, philosopher, and the bestselling author of Sapiens: A Brief History of Humankind, Homo Deus: A Brief History of Tomorrow and 21 Lessons for the 21st Century.

If you managed to get through the second of the two, I’m VERY impressed.

I wanted to add a bullet point I left out of my earlier post. At the time, I hadn't validated the data ingestion model for the primary data sources. These include datasets from the House and the Senate, as well as FEC Data, data from the Federal Record, census data, and presidential documents, including Executive Orders, proclamations, etc. A couple of small elements are still pending responses from the Historian’s Office and the Senate webmaster, but I now have all the data points to analyse millions of records very quickly.

The section that was missing from my last post was:

• The Civic Record: This space is anchored by the Federal Legislator dataset, a system-hosted list of members of the House and the Senate. This will update as needed, with a data refresh schedule attached. Here, users will have a wealth of data at their fingertips, from Member Contact details to registered interests and FEC records. You can also view bills and track their progress, including bills that die before they are ever voted on. You can also see when members departed from Party lines, along with the voting tallies on all votes, with Members listed for and against, etc.

This isn’t on the drawing board; it already has live data, and the presentation layer has been tested and debugged. I have plumbed in some baseline filters and am working on identifying where lobby money influences votes, etc. There is still some refinement work to do, but it's more at the reporting level than debugging functionality. This is the engine room at the heart of the system, and Claude and I are developing a roadmap to extend it to state government too, so we can look at influence across both federal and state government.

Another element in the design stages is monitoring the narrative each party pushes, measured against media coverage vs the facts. The system is being trained to recognise and assess patterns. It WILL take time to fine-tune the analysis engine, but I’m tuned in to what’s required.

I think this will be the most comprehensive public-focused civic hub available. I’ll do another update Sunday evening to bring you up to speed on the other elements and provide a more complete picture. I'll also note that Claude is plugged into these posts about the project; each day before we start work, it will collect feedback, anonymise it, and feed it to me so we can discuss. It collects NO personal data; it’s merely interested in assessing and collating feedback, and only for posts relating to the project. This is all designed to streamline the workflow.

Before I conclude, I want to just expand on The Frontier AI Lab. As I stated in my last piece, this will be an internal, walled-off, and secure area to explore the power of AI frontier models in terms of what they can deliver, both in terms of research and, more specifically, in the delivery of information. I’m pretty sure this will leave some gobsmacked by what it can do. The purpose of the lab is to let you share access to the same lab I’ll be using for testing as I validate Gemini AI's output. The objective is for me to fine-tune how best to prompt Gemini to get the best primary source-driven output. I have specific data stored in notebooks that I will use to test accuracy and adherence to the Truth Matters ethical engine. I want people to have hands-on experience using AI in a controlled space. Your experience will also help me fine-tune the environment.

Finally, don’t worry if you’re a Spanish speaker or have accessibility needs; I’ve got your back. The whole site will be multilingual and geo-aware too, so it will eventually deliver data specific to your state. All you need to do is allow the system to identify your location. You control every aspect of the site through the permissions you approve. If you subscribe, it will save bookmarks for any data you want to return to later.

In conclusion, I said this was a hill I was prepared to die on, and this is testament to that commitment. I have produced a supplementary post that will not be sent to your email; it is simply a record of my conversation with Gemini AI after I ran it through its convincing analysis of Project 2025. I think you might find it highly enlightening. Tomorrow, I will also be releasing an audio podcast on Substack. It’s me talking you through my journey from my first involvement with political commentary through to the vision for this project. I wanted to tell it in my own words, so keep an eye out. Until then, thank you so much for giving me some of your time today; I really do appreciate it.

You can read it by clicking the button below

If you would like to support my work but can’t commit to a paid subscription, you can now make a one-off donation by clicking the image below. Your donation helps cover research costs and keeps my independent reporting FREE for all. It also now helps support the Truth Matters Project.