Joep Meindertsma

Joep Meindertsma is the Utrecht-based Dutch software entrepreneur who founded PauseAI in Utrecht in May 2023 and now runs the organisation as Founder and Board Chair following the late-2025 handover of the CEO seat to Maxime Fournes, and is the corpus's on-record Dutch and continental-European voice on frontier-AI safety, the case for a treaty-backed pause on training the most powerful general AI systems, and the difficult psychology of confronting existential risk (see Person entry). He is tracked here as a Voice because his sustained public output — the PauseAI flagship essay "The Difficult Psychology of Existential Risk" running as the named-author treatment of the cognitive-dissonance, normalcy-bias, and ridicule-denial-fear progression that conditions audience responses to AI-x-risk warnings; the recurring podcast-interview register at ClearerThinking, Consistently Candid, For Humanity, and The Human Survival Project; the March 2024 Pakhuis de Zwijger / Existential Risk Observatory AI Safety Meetup; the Corporate Crime Reporter July 2025 interview; and his PauseAI Substack September 2024 conversation with Swedish PauseAI activist Chris Gerrby — carries the working argument that frontier-AI training above a defined capability or compute threshold is best paused now by treaty and compute governance because the systems are at present provably unsafe, that the international pause is reversible while the alternative arrival of artificial superintelligence is not, and that the substantive case for pause is owed to a public-organiser register operating outside the AI-lab and AI-safety-think-tank surface rather than to the policy-insider surfaces alone.

The Voice anchors three movement-area registers that the corpus's voices slice had previously left empty.

• The first Netherlands and continental-European voice anchor on AI safety. The corpus's voices slice carried no Netherlands voice before this entry and no continental-European voice on AI safety / frontier-AI governance — despite PauseAI being in corpus as a Utrecht-founded international federation with chapters across the Netherlands, France, Germany, Italy, and Spain, and despite the Dutch and wider continental-European existential-risk-discourse field running as a distinct sub-register inside the wider international AI-safety argument. The Person side has Meindertsma in corpus through person-joep-meindertsma; the Voice side now anchors the Dutch-founder named-author treatment of the pause case running through PauseAI's flagship policy and psychology essays, the Pakhuis de Zwijger Amsterdam talk, and the recurring podcast-interview register through which the Utrecht-founded pause movement carries into the international AI-safety field.
• The pause-movement Voice anchor outside the UK-and-US axis. The corpus's existing AI-safety Voice anchors run through the UK-and-US axis — Andrea Miotti on the ControlAI policy-and-parliamentary-track register from London — leaving the named pause-movement organiser register untouched. This Voice now anchors the parallel pause-movement register that runs through PauseAI's international federation, complementing the inside-game treaty-and-parliamentary surface that ControlAI / Miotti carry with the public-protest, mass-petition, and grassroots-mobilisation surface PauseAI carries (the May 2024 thirteen-country international protest, the November 2024 international action, the Paris AI Action Summit February 2025 mobilisation, the 2025 Google DeepMind London protest). The two Voices give the corpus the matched pair of AI-safety adult-public anchors — the inside-game policy-and-parliamentary register and the outside-game pause-movement register — sitting beside each other inside an organisational pair already linked in corpus through org-pauseai.related_orgs: [org-controlai].
• The software-entrepreneur-turned-organiser sub-type rooted in the Dutch civic-tech field. Structurally distinct from the corpus's existing voice anchors on litigators (Cori Crider, Mercy Mutemi, Daniel Motaung), framework authors (Sasha Costanza-Chock, Joy Buolamwini), lawyer-founder-and-columnists (Apar Gupta), the lawyer-organiser-and-multilateral-body register (Nighat Dad), the convener-and-essayist register (Mohamad Najem), the coalition-coordinator-and-podcast-host register (Felicia Anthonio), the policy-analyst-and-essayist register (Marwa Fatafta), and the policy-and-parliamentary-track register (Andrea Miotti) — Meindertsma's distinctive register is the software-entrepreneur-turned-organiser whose Atomic Data and AtomicServer open-source work on GitHub and prior career as CEO of his Utrecht software company sit alongside the founding and Board Chair leadership of an international pause-movement federation. The entrepreneur side anchors the substantive technical credibility on which his named commentary on frontier-AI capability rests; the organiser side anchors the named-byline grassroots-mobilisation register through which the pause case carries into the public field; the essayist side anchors the psychology-of-x-risk treatment that lifts the pause case from a policy-technical argument into a cognitive-dissonance-and-normalcy-bias argument addressed to the wider audience.

Public output and venues

Meindertsma's public-facing work runs through four overlapping channels.

• The PauseAI named-author essay register on the psychology of existential risk. The headline anchor of Meindertsma's authored public-output register is the PauseAI flagship essay The Difficult Psychology of Existential Risk, which he authored and promoted through his own LinkedIn channel in August 2023. The essay organises the psychological barriers to taking AI extinction risk seriously into four categories — difficult to bring up, difficult to believe, difficult to understand, and difficult to act on — and anchors the substantive case that AI-x-risk warnings conflict with foundational beliefs about technology being controllable and about leadership competence to ensure safety, producing cognitive dissonance that audiences resolve by ridicule, denial, and disbelief before fear eventually sets in. The essay frames the ~80% normalcy-bias rate observed in disaster research as the predictable structural drag on civil-society uptake of x-risk warnings, and is the substantive psychology-of-x-risk argument PauseAI's organising approach turns on.
• The named podcast-interview register across the international AI-safety podcast field. Meindertsma's sustained named podcast-interview register runs through the ClearerThinking show with Spencer Greenberg (Episode 207, 25 April 2024), the Consistently Candid podcast with Sarah Hastings-Woodhouse (Episode 5, 22 February 2024) on founding PauseAI and strategies for communicating AI risk, the For Humanity AI Safety Podcast "Pause AI or Die" episode with John Sherman (Episode 14, February 2024), and The Human Survival Project "Building the AI Pause Button" conversation (8 February 2024). The ClearerThinking episode anchors the substantive policy-and-capability arguments — that pausing buys time to develop provably safe AI, that current frontier systems are provably unsafe via persistent jailbreak vulnerability, that compute governance routed through TSMC and ASML answers the "bad actors will continue" objection, and that the catastrophic-risk strands include cybersecurity (AI finding zero-day exploits at scale), biological hazards (AI-assisted bioweapon creation), and control loss (systems pursuing objectives without human shutdown capability). The Consistently Candid episode anchors the emotional-experience-of-x-risk register that runs as the personal-side counterpart to the policy-side ClearerThinking case.
• The named on-record press-interview register through Corporate Crime Reporter and the Dutch existential-risk-discourse field. Meindertsma's named on-record press-interview register runs the international press circulation of the substantive AGI-to-ASI progression argument and the pause case. The Corporate Crime Reporter July 2025 interview (Volume 39, Issue 29(12), 21 July 2025) anchors the substantive AGI / ASI definitional register, the deployment of the Yudkowsky anthill analogy (humans do not plan to kill ants but wipe out anthills when building roads, and a similar resource-competition dynamic could displace humans once they are no longer the smartest species on the planet), and the airplane-test-flight risk-comparison framing addressed to AI researchers' aggregated ~14% probability of catastrophic outcomes. The March 2024 AI Safety Meetup at Pakhuis de Zwijger — hosted by the Amsterdam-based Existential Risk Observatory and framed as "what is it like, campaigning for AI Safety?" — anchors the Dutch-civil-society audience register through which Meindertsma carries the pause case into the Amsterdam existential-risk-discourse field.
• The interviewer-and-host register inside the PauseAI Substack newsroom. Beyond the named-author-and-interviewee surfaces, Meindertsma also runs an interviewer-and-host register inside the PauseAI Substack, where his September 2024 long-form interview with Swedish PauseAI activist Chris Gerrby on travelling-the-world-to-pause-AI activism anchors the named-byline conversation register through which the organisation profiles its own field of activists. The interview is also the on-record source of Meindertsma's own emotional-posture framing — his named admission of "dark thoughts about the future" sitting alongside his self-description as fitting "all the checkboxes of a really big optimist" — that anchors the personal side of the public-output register.

Signature framings

Three formulations recur across Meindertsma's public output and have done the most to install his register into the international AI-safety, frontier-AI-governance, and pause-movement field.

• "Currently provably not safe" — the provably-unsafe / provably-safe framing. Meindertsma's ClearerThinking framing of the central capability argument — that contemporary frontier AI systems are "provably not safe" because jailbreak vulnerabilities persist across releases and across the safety-fine-tuning state of the art, and that the pause is the policy correlate of the proposition that systems can be released only when they can be provably safe — is the most condensed single articulation of the burden-shifting case the corpus's pause-movement voice register turns on. The framing anchors PauseAI's policy proposal for an international AI-safety agency, a treaty pausing training above a defined capability-or-compute threshold, and the compute-governance route through TSMC and ASML, and runs directly into the substantive case made by the wider pause-movement field.
• The Yudkowsky anthill analogy — resource-competition extinction framing. Meindertsma's deployment of Eliezer Yudkowsky's anthill analogy — that humans do not plan to kill ants but routinely destroy anthills when building roads, and that a comparable resource-competition dynamic could displace humans once an artificial superintelligence is no longer competing with the smartest species on the planet — anchors the substantive proposition that the failure mode of artificial superintelligence is not malevolence but indifference under resource competition, displacing the "evil AI" framing of the popular AI-safety conversation with a substantive economic-and-ecological framing. The framing carries directly into the substantive case made in Meindertsma's named-byline press-interview register and his podcast-interview public-speaking surface.
• "Ridicule, denial and disbelief, then fear" — the psychology-of-x-risk progression framing. Meindertsma's authored treatment in The Difficult Psychology of Existential Risk is the substantive proposition that displaces the assumption that civil-society uptake of AI-x-risk warnings will run on the merits of the technical case alone, framing the audience response instead as a predictable cognitive-dissonance progression — ridicule first, denial and disbelief second, fear only later after extended reflection — that is conditioned by foundational beliefs about technology being controllable and leadership being competent to ensure safety, by the ~80% normalcy-bias rate observed in disaster research, and by the appeal of comforting alternative explanations (conspiracy framings, cult-framings) as psychological shields. The framing carries directly through into PauseAI's organising approach, where the named recognition of the cognitive-dissonance progression as the structural drag on uptake conditions the choice of public-protest, mass-petition, and sustained-conversation tactics rather than insider-only policy work.

Organisational vehicle

Meindertsma's public output runs primarily through PauseAI, where he serves as Founder and Board Chair of PauseAI Global after the late-2025 handover of the CEO role to Maxime Fournes, and through the PauseAI federation of national chapters (PauseAI US under Holly Elmore, PauseAI UK under Joseph Miller, PauseIA France under Clémence Peyrot, PauseAI Canada under Jeremy Eliosoff, PauseAI Germany, and chapters across Australia, Spain, Italy, and the Netherlands). The organisational vehicle carries the substantive connective tissue between the founding Utrecht software-entrepreneur-turned-organiser surface, the international federation's volunteer-led national-chapter infrastructure, and the international AI-safety field — anchored in corpus through the related-orgs link to ControlAI on the policy-and-parliamentary side — and is the substantive reason his Voice anchors the corpus's first pause-movement and first Netherlands voice from inside the international federation's grassroots-mobilisation infrastructure rather than from the policy-insider or AI-lab-affiliated surfaces.