Anthony Albanese says he has spoken to the Open AI chief executive to express his concern about the incident and the length of time it took the tech company to inform the government of the breach.

Can AI systems be aligned to human values? The popularization of large language models (LLMs) and multi-modal foundation models has seen a rise in harms spanning from toxic speech and hallucinations to AI agents executing unauthorized actions. Within the field of AI safety, these harmful instances are often framed as the alignment problem, or of models being misaligned with human values. Researchers have responded by pursuing applied and theoretical AI value alignment efforts, often without specifying what they mean by human values. How does the field of AI value alignment conceive of human values? How are these conceptions of values technically operationalized and evaluated? What does the emergent theory of value from this field signify for the future of AI? We annotated 94 value alignment research papers to discern their implicit theory of values in AI. The majority do not define values, relying heavily on preferences as a stand in that runs the risk of reducing complex culturally situated concepts down to binary choices. As researchers dispense with using human annotators for model training and evaluation, turning instead to synthetic data and autorater approaches to aligning and evaluating models, we identify the potential to close off alternative methods for contesting and enacting values in foundation models. In making AI value alignments philosophical commitments explicit, we seek to bring great specificity and under explored perspectives in the debate on whether and how AI can address human values.

## Work with us * * * ### ICCL Enforce is hiring ### Enforce Artificial Intelligence Expert * * * Enforce, a unit of the Irish Council for Civil Liberties (ICCL), is hiring an expert in law / economics / computer science, or other relevant domain, to help ensure supervision and enforcement of human rights in the context of A.I. and automated decision making systems. We will assess the applications on a rolling basis until 5pm Dublin time on **15 April 2026**. Please apply early. **Duration** : A full-time role is funded till 31 August 2027, with the possibility to renew based on funding extension. **Starting Salary** : €65,000 — €100,000 per annum depending on prior experience and local social charges. **Start Date:** As soon as possible, preferably before 4 May 2026. **Location** : Remotely in the EU. ICCL Enforce is based in Dublin. You may have to travel occasionally, particularly to Brussels. **What you will do ** Enforce investigates, advocates, and litigates to protect people and their rights. In Enforce, you will be expected to * engage with senior stakeholders and experts in key jurisdictions; * influence the implementation and enforcement of law and policy, including implementing acts of the EU AI Act; * investigate and expose harmful use of AI; * track technology and market developments; and * most particularly, track policy developments. You will be expected to work independently, potentially leading projects, with the support of the Director of Enforce, to whom you will report. **Qualifications** REQUIRED * Expertise in at least one of the following aspects of AI: legal and regulatory, and/or economic (including markets and labour), and/or computer science. * Excellent written and spoken English. * Ability to write concisely and communicate clearly. * Ability to work with colleagues and other stakeholders without expertise in AI/automated decision making systems. * Legal right to work in an EU country. * Capacity to learn and to take advantage of feedback. DESIRED * Strong academic or industry or policy record. You may have law degree, or MSc/PhD in economics, statistics, computer science, engineering, or a related field. * Familiarity with EU digital law, particularly the AI Act and GDPR. * Record of communication with and influence on stakeholders including experts, law makers, regulators, standards bodies, industry, and media. * Existing relationships with stakeholders. * Existing relationships with philanthropic foundations. ** To apply** Send the following documents to info@iccl.ie with 'Enforce A.I. specialist - Application' in the subject line of the email before 5pm Dublin time on 15 April 2026: 1. CV of maximum three pages 2. Describe in less than a page the first project you would like to pursue in this role. Describe how you would approach the project based on your strengths and the impact of the project. If you used generative AI to aid in your application, explain why in a separate document along with all the prompts and tools you used. **About Enforce** Enforce is a unit of the Irish Council for Civil Liberties. Ireland’s unique responsibility for monitoring human rights on large digital platforms give us an international focus. We share technical expertise with legislators in strategic jurisdictions, and we investigate tech, industry practice, and markets. Our expertise supports other civil society organisations, too. We also take bad actors to court in key jurisdictions. Learn more at https://iccl.ie/enforce/

_**9 March 2026 - Report finds that PETs often fall short on their promise of safeguarding privacy without limiting data use.**_ ICCL and Maynooth University have published a new report by ICCL's Dr Kris Shrishak and Dr Maria Helen Murphy, Maynooth University School of Law on the subject of Privacy-Enhancing Technologies (PETs). PETs are often presented as tools that enable the free flow of data while protecting personal information. In practice, however, PETs often fall short on their promise of safeguarding privacy without limiting data use. When privacy is reduced to a narrow conception of confidentiality, the tools marketed to enhance privacy risk undermining protection and entrenching existing concentrations of power. Drawing on two high-profile attempts at real-world deployment that were ultimately discontinued, this report shows how current approaches to PETs can obscure important risks, and that PETs must be understood as one component of a wider, rights-based governance framework rather than a technical fix for privacy problems. Crucially, broad engagement with diverse interdisciplinary stakeholders is an essential part of the process. In a time of legislative flux, where ‘simplification’ in pursuit of innovation is prominent in policy discussions, the limitations of PETs as a catch-all ‘solution’ must be accurately understood. This is especially the case as ongoing debates on the very ‘notion of personal data’, as reflected in the EU Digital Omnibus proposals, have the potential to further emphasise technical protections based on concepts of identifiability at the expense of full consideration of privacy harms. * * * Download the report (PDF)

176 Posts, 525 Following, 891 Followers · Media in challenging environments