Primary sources before opinions.

Spain's 2024-2025 acceleration plan, adding EUR1.5B to EUR600M already mobilized and organizing the agenda around AI capacity, adoption, and humanistic governance.

Readable official summary of the 2024 strategy, including ALIA, supercomputing, public and private-sector adoption, SMEs, and AESIA's ethical-governance role.

The original national AI roadmap from 2020, still essential for understanding how Spain moved from strategy to AESIA, sandboxes, data infrastructure, and adoption programs.

The broad digital transformation agenda that situates AI inside infrastructure, public-sector modernization, SME digitalization, skills, and digital rights.

The citizen-facing technology innovation lab for public administration, intended to test AI, data analytics, blockchain, and new service models.

The SME advisory program for specialized digital consulting, including AI and data use cases for companies between 10 and fewer than 250 employees.

The mass digitalization grant program for small firms and self-employed workers, relevant because AI products and analytics now sit inside Spain's broader SME adoption policy.

The bill published by Congress on 12 Jun 2026: supervisory authorities, sanctions, public-sector AI inventory, AI delegates and AESIA-operated sandboxes. The latest amendment deadline published by Congress is 9 Sep 2026 at 18:00.

AESIA's institutional site, useful for tracking the supervisor's public agenda, lab work, guidance, and future national enforcement role.

The legal basis for Spain's AI supervisory agency, including its supervision, advice, awareness, training, testing, coordination, and inspection functions.

Spain's rider-law amendment to the Workers' Statute, requiring worker representatives to be informed about algorithmic or AI systems affecting work conditions, access, retention, or profiling.

Spain's digital-rights charter, a policy reference for human-centered digitalization and rights language around automated decision-making and AI.

A short explainer on the first ALIA models in Spanish and co-official languages, with useful context on the model family, methodology, datasets, and the infrastructure value behind public AI rather than consumer-model rankings.

Core adoption benchmark: 11.4% of Spanish firms with 10+ employees used AI in 2024, with large-firm adoption much higher and citizen GenAI use now mainstream.

INE's official business-technology survey is the statistical base behind many AI, big-data, cloud, and adoption indicators in Spain.

Spain's public-private venture-capital instrument for scaleups and funds in AI, cloud, IoT, cybersecurity, NLP, biometrics, and other high-impact digital technologies.

The training layer under Spain Digital: citizens, workers, public employees, vocational education, specialists, and gender gap reduction.

A Spain Digital measure to attract and retain AI talent, connecting researchers, companies, investment opportunities, scholarships, and job offers.

A useful pulse check on SME advisory uptake, noting thousands of grants and the role of AI, data analytics, cybersecurity, and business-performance advice.

A 2025 progress update on Spain's digital-skills plan, including public employees, teachers, citizens, professional associations, and AI-applied work training.

AESIA's 2026 public-interest agenda, starting with working lines on women and AI, disinformation and AI, minors and AI, and labor-market impacts.

A 2026 public-facing campaign for young people, families, and educators on AI use, digital safety, generated-content recognition, parental controls, and responsible classroom adoption.

INCIBE's citizen-facing guide to AI risks, including malicious use, phishing, malware generation, deepfakes, false profiles, and practical detection cues.

A practical deepfake primer from INCIBE, useful for media literacy, minors, fraud prevention, and democratic-resilience conversations.

A RETECH/INCIBE project to define control frameworks and risk-management methodology for AI systems against regulatory and cybersecurity expectations.

The broader security frame for AI-enabled threats, digital resilience, incident response, critical infrastructure, and Spain's national-security coordination.

AEPD's 2026 privacy recommendations for citizens using AI tools, including avoiding sensitive disclosures and understanding image and personal-data risks.

Spain's long-range AI foresight study: private-sector modernization, welfare state, sustainability, security and defence, and inequality through a 2040 horizon.

AESIA's framing of the foresight report, including the priority line to build AI verification and auditing tools through the supervisor.

A late-2025 marker for Spain's public AI trajectory: a larger language model adapted to Spain's linguistic and cultural context. The sober reading is infrastructure first: benchmark performance should be tracked, but it is not the whole public-value test.

The flagship public-language-model project: open AI resources for Spanish, Catalan/Valencian, Basque, and Galician. Read it as a multi-institution public data and language-infrastructure program, not as a claim to beat commercial frontier models.

AESIA's 29 May 2026 note links Spain's supervisor to the European AI Office working groups behind the transparency and high-risk draft guidance, a concrete signal of Spain's role in the AI Act implementation layer.

Spain's pre-AI-Act sandbox, created to rehearse compliance for high-risk AI, general-purpose systems, and foundation-model-like systems before full EU enforcement.

The best entry point for Spanish data-protection guidance on AI, including agentic AI, GenAI use, audits, privacy-by-design, and DPIA-adjacent practice.

A 2026 AEPD guide on AI agents, autonomy, environmental interaction, and measures controllers and processors can use to manage data-protection risks.

The Barcelona Supercomputing Center's AI Factory program: compute access, training, innovation support, and services for startups, SMEs, corporates, academia, and public institutions.

The 2026 contract for the AI Factory upgrade, adding training and inference partitions, high-performance AI storage, and specialized software for industry, science, SMEs, startups, and public administration.

Official Spain Digital note on the Galicia-based AI Factory, which broadens Spain's compute story beyond Barcelona by linking CESGA, health, life sciences, public-interest AI, and EuroHPC capacity.

Practical access page for the BSC AI Factory's AI-ready compute resources, including the AI-optimized partition and support model.

Entry point to Spain's data-spaces agenda, including CRED and the plan to make sectoral data spaces part of AI-ready data infrastructure.

La Moncloa's Spanish-source readout of the 7 May 2026 trilogue agreement, including Spain's role, the proposed prohibited practices, and the pending formal ratification path.

The European Commission selection of Spain's second AI Factory, connecting sovereign compute capacity to health, science, public-interest AI, and regional infrastructure.

The operational clock for Spain, read alongside Regulation (EU) 2026/1744: Article 50 transparency applies from 2 Aug 2026, while the main Annex III and Annex I high-risk requirements now apply later.

A compact map of the EU's excellence-and-trust approach, linking the AI Act with investment, AI factories, data, and adoption policy.

Frequently asked questions on AI Act scope, implementation, governance, enforcement, GPAI, and Digital Omnibus uncertainty.

The legal source text. Read Article 5 for prohibitions, Article 50 for transparency, GPAI provisions, high-risk obligations, sandboxes, and market-surveillance architecture.

The binding amendment to the AI Act: Annex III high-risk requirements move to 2 Dec 2027, product-related Annex I requirements to 2 Aug 2028, with new prohibitions and a limited Article 50 transition.

The Commission's final Article 50 guidance on AI-interaction notices, machine-readable marking, deepfake labels, public-interest synthetic publications, and biometric or emotion-recognition disclosures.

The Commission's optional icon set for certain AI-generated or manipulated content. The icons can support Article 50 disclosure, but using them is not mandatory and does not establish compliance by itself.

The 19 May 2026 draft guidance and practical examples for deciding whether an AI system is high-risk under Article 6, Annex I, or Annex III of the AI Act, with consultation open until 23 June 2026.

EuroHPC's official announcement naming Barcelona's BSC AI Factory as one of the first European AI Factories and linking Spain's capacity to the wider EU compute strategy.

The operational EU entry point for systemic-risk GPAI: all GPAI providers face documentation, copyright, and training-content duties; systemic-risk models add notification, risk mitigation, incident reporting, and cybersecurity obligations.

The EU's practical compliance instrument for GPAI providers, with Transparency, Copyright, and Safety and Security chapters. Watch it closely: Nuria Oliver chaired the Transparency working group, while the Safety and Security chapter is central for systemic-risk models.

The EU's 2025 sectoral adoption strategy, relevant to Spain because it links public-sector AI, SMEs, European AI solutions, open source, data, and AI factories.

The 2026 EuroHPC amendment enabling AI gigafactories, a next-step compute-policy issue for Spain as Europe scales from factories to world-class model infrastructure.

EuroHPC's explainer on large-scale AI facilities for next-generation model development, inference, storage, secure cloud environments, and specialized support services.

The Commission and EuroHPC competition to support a European frontier model with EuroHPC compute capacity, open outputs, and strategic autonomy goals.

The 2026 progress marker for Europe's shift from regulation to competitiveness: AI factories, data, skills, adoption, and strategic capacity.

A concise baseline for trustworthy AI policy: inclusive growth, human-centered values, transparency, robustness, security, safety, and accountability.

The global rights-and-values reference for public AI governance, covering proportionality, safety, fairness, sustainability, privacy, and human oversight.

The Council of Europe treaty frame for democracy, human rights, and rule-of-law protections in AI systems.

Spain's 2026 link to global AI governance: a UN lab based in Valencia to support AI governance work under the UN Office for Digital and Emerging Technologies.

The best current scientific baseline for advanced general-purpose AI risk: capabilities, misuse, loss-of-control debates, risk-management techniques, and evidence gaps, backed by 30+ countries and international organizations.

The frontier-model evaluation network to compare with Spain's architecture: Spain participates in international safety-report processes and has AESIA for AI Act supervision, but is not publicly listed as a national AI Safety Institute member.

The 2023 international declaration, with Spain listed among represented countries, that frames frontier AI risks as international and calls for shared science, evaluation, transparency, and public-sector capability.

A practical risk-management bridge for public and private organizations: govern, map, measure, and manage AI risks, with a 2024 generative-AI profile and 2026 critical-infrastructure work in progress.

A non-technical course useful for policymakers, journalists, and civil servants who need a shared vocabulary before reading legal or technical sources.

A survey of 2,778 AI researchers: the wiki reports 5%, 10%, and 5% median responses across three extinction or severe-disempowerment phrasings, plus 57.8% giving at least 5% odds to extremely bad HLMI impacts.

METR's task-horizon work makes autonomy legible for policy: the March 2025 paper reports roughly seven-month doubling over six years, while the newer live estimates are faster and should be read separately.

METR's live autonomy dashboard: 50% and 80% reliability time horizons for public frontier agents. The 8 May 2026 update adds Claude Mythos Preview (early), about a 17.4-hour p50 and 3.1-hour p80 estimate, plus an explicit warning that measurements above 16 hours are unreliable with the current suite.

A policy-useful map of frontier capability benchmarks across reasoning, coding, agents, multimodality, and other domains, useful for avoiding single-score model hype.

Epoch AI's expert-level mathematics benchmark, designed to test advanced reasoning beyond saturated standard exams and to support more serious capability measurement.

The Nature paper for Humanity's Last Exam, a 2,500-question expert benchmark across subjects; useful because it shows both harder evaluation design and continuing uncertainty.

Epoch AI's terminal-task benchmark for agentic command-line work, a useful proxy for software operations, debugging, and tool-using capability.

A benchmark for multimodal agents doing open-ended tasks in real computer environments; relevant to public-sector AI because autonomy increasingly means operating software, not just generating text.

A concrete evaluation toolkit for dangerous autonomous capabilities, including task-suite guidance and an example protocol for measuring agentic frontier-model risk.

A frontier-lab primary source on severe-risk categories, including cybersecurity, CBRN, persuasion, autonomy, evaluations, red-teaming, reporting, and security controls.

A 28 May 2026 frontier-lab governance source mapping safety and security practices to emerging legal requirements, including the EU AI Act GPAI Code of Practice track, cyber, CBRN, manipulation, loss-of-control, reporting, security, and incident response.

A frontier-lab primary source on Anthropic's RSP v3.0: it replaces the earlier hard-pause style trigger with public Frontier Safety Roadmaps, goals, risk reports, safeguards, evaluations, and security requirements.

DeepMind's updated frontier-safety framework adds tracked capability levels and a risk-management process for severe risks from advanced models, useful for comparing lab governance with public oversight.

An influential 2023 public statement that helped push extinction-risk language into mainstream debate; include it as a signal of expert concern, not as a substitute for evidence, law, or institutional design.