Success may include stronger company safeguards, more rigorous risk assessment and mitigation, greater transparency, effective regulatory oversight and clearer accountability for decisions concerning powerful AI systems. The wider aim is to ensure that safety practices develop alongside technological capabilities.

JUST is independent from the AI industry. We are not an AI company, an industry association, or a lobbying organisation acting on behalf of commercial developers. AI companies do not appoint our board, approve our research, determine our strategy, or control our publications.

‍

As explained above, some of our funders have connections with the AI industry or the wider AI safety community. We disclose those connections for transparency reasons, but they confer no control over our work.

‍

We select our work and reach our conclusions independently, based on the evidence, the applicable law, and our mission to protect the public from serious harm.

We prioritise constructive engagement informed by legal, technical and policy expertise. Depending on the circumstances, we may engage companies directly, work with public authorities and civil-society partners, seek greater transparency or pursue regulatory complaints and other proportionate legal action.

We seek proportionate safeguards. These may include robust model evaluation, risk assessment and mitigation, cybersecurity measures, incident reporting, meaningful documentation and transparency, and governance arrangements that give serious consideration to foreseeable harms.

AI providers make decisions that determine what advanced systems can do, how they are tested, who can access them and which safeguards accompany their deployment. Their choices therefore have direct consequences for how systemic and potentially catastrophic risks are managed.

The appropriate measures depend on the company and its position in the supply chain. They may include stronger AI-risk assessment, board-level oversight, meaningful disclosure, integration of technological risks into human-rights and sustainability processes, and responsible safety measures within commercial relationships.

We prioritise constructive dialogue informed by legal, technical and policy expertise. Depending on the context, we may engage boards, management, shareholders and other stakeholders, independently or with partners. Legal action is considered only where warranted by the evidence, the company’s response and the potential public benefit.

Not automatically or without limit. The responsibilities of infrastructure companies, i.e. companies that provide critical infrastructure to facilitate transformative technologies, depend on factors including their knowledge, influence, legal obligations and practical ability to mitigate risk. JUST examines when risks related to the business of these companies (or the ones that they supply) are sufficiently serious and foreseeable to warrant assessment, disclosure or proportionate action.

Certain infrastructure companies occupy positions that are difficult to replace. Their commercial relationships may extend across equipment manufacturers, chipmakers, compute providers and AI developers, giving them unusual visibility and influence throughout the technology supply chain.

The critical supply chain, or value chain, includes the specialised technologies, equipment and services on which transformative technologies depend. In artificial intelligence, this includes the infrastructure supply chains for semiconductors and compute that enable the development and operation of the most capable AI systems. In biotechnology, it includes infrastructure such as DNA-synthesis services and advanced laboratory capabilities.

The EU AI Act provides an important framework for holding AI providers accountable, particularly where general-purpose AI models may pose systemic risks. JUST examines how its requirements concerning evaluation, risk management, mitigation and transparency can be implemented and enforced effectively, including by civil society organizations like JUST.

Both may be appropriate. We prioritise constructive engagement and collaboration where these can produce meaningful safeguards. Where an inadequate response leaves serious risks unaddressed, we may pursue transparency procedures, regulatory complaints or other proportionate legal action.

We focus on decisions, or the lack thereof, and frameworks with the potential to materially reduce risk. These include technology regulation and supervision, export licensing, end-use controls, regulatory enforcement, risk assessments and transparency around government decision-making.

European governments regulate markets, control exports, supervise compliance and enforce legal safeguards. Their decisions can influence the development and deployment of transformative technologies both within Europe and across international technology supply chains.

Effective Altruism is a broad and decentralised movement concerned with using evidence and careful reasoning to determine how money, time, and other resources can do the most good. It includes people and organisations with widely differing priorities, including global health, animal welfare, poverty reduction, pandemic preparedness, and risks from advanced technologies. It is not a single organisation, and it does not have a central authority that determines the positions of everyone associated with it.

‍

Some of JUST’s funders are associated with the Effective Altruism community or support causes that are prominent within it. The Survival and Flourishing Fund, for example, has funded Effective Altruism organisations and infrastructure, as well as work on risks from advanced AI. [The Future of Life Institute and] Macroscopic Ventures also operate within the wider ecosystem of organisations concerned with long-term and potentially catastrophic risks.

‍

JUST is not governed by the Effective Altruism movement and does not represent it. We do not adopt positions because they are popular within that community, and we do not require our staff, board members, or partners to identify with it. Where our work overlaps with concerns raised by Effective Altruism, particularly the prevention of large-scale harm, that overlap reflects a shared subject of concern rather than institutional control or automatic agreement.

‍

We remain free to disagree with Effective Altruist organisations, funders, AI companies, and other actors. Our conclusions are based on our own legal research, factual assessment, and mission.

The fact that some AI executives have themselves warned about advanced AI risks does not mean that independent organisations addressing those risks are acting in the industry’s interests.

‍

JUST does not take a position because an AI company or its executives have expressed the same concern. Nor do we assume that a warning is wrong merely because it comes from industry. We examine the available evidence, the applicable law, the severity and probability of potential harm, and the consequences of both regulatory action and inaction.

Where we support a pause or another precautionary measure, we do so because our assessment indicates that it may be necessary to prevent serious harm. We also consider whether a proposed measure would genuinely reduce risk or instead primarily benefit established companies, for example by creating unnecessary barriers to entry.

‍

Our objective is not to generate publicity for AI companies or to increase their perceived importance. It is to identify proportionate and effective legal safeguards for technologies capable of causing serious harm.

JUST is an independent Dutch nonprofit foundation. Our work is currently supported by philanthropic funders, including Macroscopic Ventures and grants associated with Jaan Tallinn through the Survival and Flourishing Fund.

‍

Some of these funders are associated with the Effective Altruism community or with work on the societal risks of advanced AI. There are also specific connections between some funders and the AI industry. For example, Macroscopic Ventures has publicly disclosed investments in Anthropic. Jaan Tallinn has both philanthropic and investment activities relating to AI. The Survival and Flourishing Fund also supports several organisations associated with Effective Altruism.

‍

We disclose these connections because transparency about funding matters. Funding can create actual or perceived conflicts of interest, particularly where funders have relationships with organisations affected by our work. Disclosing relevant relationships allows others to assess our independence on the basis of accurate information.

‍

Our funding does not give our donors a say over our strategy, research projects, legal analysis, advocacy, or public communications. Our funders do not approve our positions or determine which companies, technologies, or practices we examine. We decide independently how to use our funding in pursuit of our mission.

‍

We do not serve the commercial interests of AI developers, our funders, or any political or intellectual movement. Our purpose is to serve the public interest by examining how law can reduce the risk of large-scale harm from powerful technologies. We seek to demonstrate our independence through transparent funding disclosures and rigorous, evidence-based legal work.

‍

We assess each opportunity for potential impact, strategic fit, evidentiary strength, legal merit, and possible unintended negative consequences. Our analysis is also tested with external legal, technical and/or policy experts before we make a decision on how to proceed. We proceed only where the facts, legal basis and expected public benefit support action.

Such steps include stronger corporate understanding of AI-related risks, improved governance and disclosure, practical safeguards across the value chain and durable commitments reached through constructive dialogue. Individual engagements can also generate knowledge and approaches that inform responsible practice across the wider infrastructure sector. This is what JUST works towards.

JUST does not begin from the assumption that broad restrictions or bans are the right answer to the challenges posed by the current unregulated development of transformative technologies. We seek targeted, proportionate safeguards that reduce catastrophic risks of increasingly powerful systems to be developed and deployed, while taking into account beneficial uses, technical feasibility and the strategic and geopolitical position of the companies involved.

Depending on the risk and decision involved, relevant frameworks may include human-rights law, WMD non-proliferation commitments, dual-use export controls, the EU AI Act and related public-law and duties-of-care principles. We treat these frameworks as practical tools for clarifying responsibility and advancing safeguards, rather than as separate fields of academic inquiry.

We believe that their warnings need to be taken very seriously because of what is at stake. We believe that risks of the scale that generative AI and industry experts warn about should be reduced to effectively zero. The alignment problem, the risk of losing control, the potential for misuse e.g. cybersecurity and bio-attacks or accidents are well substantiated, and supported by observed behaviors of AIs in test environments as well as loss of control and cyberhacking incidents, exposing society’s critical infrastructure which run on software, such as energy, water and food supply as well as healthcare, to devastating cyber attacks. Also, recently scientists used AI to create novel viruses that proved viable among bacteria. These are not harmful for humans, but soon AI’s may be capable of modifying existing viruses or creating new ones that are. Both independent AI experts as well as the AI labs themselves warn that these risks are real, but that we don’t have reliable solutions for them and are not confident they can be fully managed at all. A fundamental problem is that no one, not even the people developing the technology, understands exactly how it works: we don’t know exactly what goes on in the artificial brain any more than we know what goes on in our brains.

Potentially catastrophic risks from advanced AI and from biotechnology. With regard to AI, amongst others, we work on preventing risks from systems capable of enabling serious biological, chemical or cyber harm, and from failures and unexpected behaviour in critical deployments, as well as  longer-term risks from autonomous systems operating outside meaningful human oversight. With regard to biotechnology, amongst others, we work on preventing risks from the convergence of AI tools with biological engineering, which lowers the barriers to designing or acquiring dangerous pathogens, and the misuse or accidental release of pathogens capable of causing a pandemic. These two increasingly overlap, which is why we work on them together. 

We are Stichting Legal Alliance for Just Technologies (“JUST”). JUST is a Dutch non-profit foundation that uses law to reduce the risk of large-scale harm from transformative technologies. Read more about our work and our leadership on our 'How we work' page.