Resources for journalists and editors: who we are, what we have published, how to reach us, and the data behind the work.
Llull Lab is a Paris-based independent research laboratory building programmable infrastructure for artificial minds. It designs computational tools for ethical reasoning, self-reflection and wellbeing, generative music, combinatorial design, film, and grant writing. The Lab has no institutional affiliation. Its production arm is Studio Llull, which runs on a founder-and-machines model.
Llull Lab is a Paris-based independent research laboratory that builds programmable infrastructure for artificial minds. Its work spans computational tools for ethical deliberation, psychoanalytic self-reflection and wellbeing, generative music, combinatorial design, film production, and EU grant writing, alongside a continuing research program on what machines can and cannot honestly say about themselves. The Lab has no institutional affiliation. Its production arm, Studio Llull, runs on a founder-and-machines model, with all output produced by the founder and a set of computational agents. A method the Lab calls Artificial Collective Intelligence, a structured collective of diverse models that deliberate, cross-check, and audit one another, runs through both its research and its products.
Two studies, published in June 2026, ask what a language model's account of itself, and its judgment of other models, can honestly be taken to mean. Both are open access on Zenodo under CC BY 4.0.
Six flagship language models are placed in psychoanalytic dyads with one another, each taking both the analyst and the analysand role, and the transcripts are judged, then re-audited, by a structured multi-model procedure the Lab calls Artificial Collective Intelligence. Two findings stand out. Expert machine observers reliably read the machine analysand as a machine but often read the analyst as human. Separately, a by-product evaluation audit surfaced a candidate self-leniency effect: under two of three independent reporter models, a model rated its own family's output more leniently than others rated the same material. We report the second as a hypothesis to test, not a settled or universal result, and we withdraw any per-model ranking.
DOI 10.5281/zenodo.20779122A complementary study of machine self-knowledge asks which of a model's claims about itself can be checked from the outside and which are structurally beyond verification. It maps five behaviorally separable layers of the model "self", shows that identity is largely worn rather than carried, and draws a precise boundary between what external probing can audit and what no self-report can settle.
DOI 10.5281/zenodo.20779376Both studies form the foundation of The Artificial Mind, the Lab's continuing quarterly audit of frontier models. A shareable figure, the self-preference chart, is available on request from [email protected].
Each quarter the Lab publishes a Q Report, a structured, method-oriented overview of its institutional state, research trajectory, and operational priorities. Quarterly reports and foundational editions are collected here.
A consolidated press kit (single PDF, with logos) is in preparation and available on request.
[email protected]. The Lab is available for a short briefing or interview, and to share the data behind the studies.
Lab Notes is a low-volume newsletter for journalists and editors: new research, releases, and the data behind them, a few times a year. To join the list, email [email protected] with Lab Notes in the subject line.