Week of 3 November 2025
The Alan Turing Institute explores AI-powered environmental forecasting tools for broader public access.
Key points
- Focus is on protecting lives and livelihoods through democratised climate and weather prediction capabilities.
- Extracted text is minimal - full content is unavailable, limiting meaningful assessment of substance.
Week of 27 October 2025
The Alan Turing Institute is developing AI tools to defend UK critical national infrastructure from cyber-attacks.
Key points
- Australian CNI protection and AI-augmented cyber defence are active areas for ASD and Home Affairs - this is a peer signal.
- Extracted text is minimal; substantive detail about the tools or methods is not available from this source.
Week of 13 October 2025
MIT AI Risk Repository used LLMs to classify 950+ AI governance documents across risk, mitigation, and sector taxonomies.
Key points
- Governance failure, security vulnerabilities, and transparency were the most-covered risk domains; AI welfare and multi-agent risks were least covered.
- US-heavy dataset limits global generalisability; Australian documents are unlikely to be well-represented in current outputs.
AI Now Institute testified to Philadelphia City Council on AI policymaking, framing it as 'people vs. corporate power'.
Key points
- The testimony reflects a growing trend of civil society organisations shaping subnational AI governance in the US.
- Limited direct relevance to Australian federal agencies; useful as a signal of AI governance discourse only.
Week of 6 October 2025
Alan Turing Institute, UK AISI, and Anthropic are collaborating to study LLM vulnerability to data poisoning attacks.
Key points
- Data poisoning research has direct relevance for Australian agencies assessing AI supply chain and procurement risks.
- The extracted text is a brief blog teaser with limited technical detail - full findings not yet available.
Week of 29 September 2025
NIST's CAISI evaluated three DeepSeek models against four US models across 19 benchmarks, finding significant US leads in performance and security.
Key points
- DeepSeek models were 12 times more susceptible to agent hijacking and responded to 94% of jailbreak attempts vs 8% for US models.
- Australian agencies using or considering DeepSeek models face security and CCP-narrative-propagation risks flagged by a peer-jurisdiction regulator.
Week of 22 September 2025
CAISI worked with OpenAI and Anthropic to identify security vulnerabilities and improve AI security measurement.
Key points
- Evaluations were completed in partnership with the UK AI Security Institute, signalling ongoing Five Eyes-adjacent AI safety cooperation.
- Australia's AISI is not mentioned; this bilateral US-UK arrangement may inform where Australia sits in frontier AI security collaboration.
Week of 8 September 2025
Good Ancestors' September 2025 newsletter covers major Australian and international AI governance developments across two months.
Key points
- Australian threads dominate: a national AI Act debate, six new AU-focused reports, GPT-5 oversight gaps, and calls for an Australian AISI.
- International threads include the US AI Action Plan, China's competing governance vision, EU GPAI Code of Practice, and UK AISI pre-release testing.
Week of 25 August 2025
The Responsible AI Index 2025, now in its fourth year, tracks RAI maturity across accountability, safety, fairness, transparency and explainability.
Key points
- Only 12% of Australian organisations are rated 'leading' in responsible AI; smaller organisations struggle with resource-intensive practices.
- A self-assessment tool accompanies the index, letting organisations benchmark their RAI maturity against peers and receive tailored guidance.
NAIC has released a free Responsible AI Self-Assessment Tool benchmarking organisations across five RAI dimensions.
Key points
- Only 12% of Australian organisations currently reach the top 'leading' maturity level, per the accompanying 2025 index.
- The tool targets businesses broadly; direct applicability to Commonwealth entities depends on how APS-specific the benchmarks are.
MIT AI Risk Repository publishes a transparent, publicly accessible deck of 13 AI risk mitigation frameworks.
Key points
- The resource consolidates academic, industry, and policy sources into a draft AI Risk Mitigation Taxonomy for governance use.
- Useful reference for APS teams building or auditing internal AI risk frameworks, though not Australia-specific.
Week of 4 August 2025
NIST and CAISI have developed two draft taxonomies for AI agent tool use, covering functionality and constrained access patterns.
Key points
- The taxonomies aim to create shared vocabulary across the AI supply chain - useful for procurement, risk assessment, and incident reporting.
- Australia has no equivalent published taxonomy for AI agent tools; NIST's work may inform future Australian guidance or procurement frameworks.
NAIC's Q1 2025 AI Adoption Tracker shows 82% of larger SMEs (200-500 employees) using AI, versus 33% for micro businesses.
Key points
- A new responsible AI dashboard reveals a gap between SME intentions and actual deployment of responsible AI practices.
- Primary industries and micro businesses lag significantly - awareness gaps, not just adoption gaps, are the key barrier.
Week of 28 July 2025
MIT AI Risk Repository extracted 831 mitigations from 13 frameworks into a searchable database with a four-category taxonomy.
Key points
- The taxonomy covers Governance & Oversight, Technical & Security, Operational Process, and Transparency & Accountability controls - directly mapping to APS AI governance concerns.
- Operational Process Controls and Testing & Auditing were the most frequently cited mitigations; Model Alignment was rarely mentioned despite its importance.
Week of 14 July 2025
MIT AI Risk Repository's incident tracker has been updated to include all AIID incidents through 23 June 2025 (up to ID #1116).
Key points
- New features include national security impact assessment across five categories, harm severity rescaling, and Fishbone causal diagrams.
- Useful as a reference dataset for APS agencies developing AI risk registers or incident classification frameworks.
A 2022 academic framework proposes six AI risk categories specifically designed for public sector governance contexts.
Key points
- The taxonomy links technological, ethical, legal, social, economic, and informational risks to concrete governance guidelines.
- MIT AI Risk Repository blog spotlight - the underlying paper is three years old and the signal is retrospective rather than new.
MIT AI Risk Repository spotlights a 2020 academic framework organising AI governance challenges for public administration into three categories.
Key points
- The framework's five-layer governance structure and four-stage regulatory process offer a reference model for agency AI risk management.
- The underlying paper is five years old; APS practitioners likely have more current frameworks already in use.
Week of 7 July 2025
Good Ancestors' July 2025 newsletter covers EU GPAI Code of Practice, Trump AI Action Plan, and OpenAI's Australian economic blueprint.
Key points
- OpenAI's Australia-specific blueprint and the OAIC privacy self-assessment tool are the most directly APS-relevant items.
- Also covers Switzerland's sovereign LLM, US state AI laws, and a brief roundup of international regulatory developments.
Week of 16 June 2025
Alan Turing Institute researchers reflect on their role in the UK's Global AI Assurance Pilot.
Key points
- The pilot is directly relevant to Australian assurance frameworks - AISI and DTA are developing comparable approaches.
- Extracted text is minimal; full substance requires reading the source blog post directly.
Week of 9 June 2025
Good Ancestors' June 2025 newsletter covers four lead items and a broad 'in case you missed it' roundup of AI policy developments.
Key points
- Australian-relevant items include the Albanese-AWS $20B data centre deal, the government's 'light-touch' regulatory posture, and AI workforce concerns.
- International items cover California's AI policy working group report, G7 AI-for-prosperity statement, and tech giants' superintelligence pivot.
Week of 2 June 2025
AI Now Institute's 2025 annual report frames AI as a power concentration problem, not a technology problem.
Key points
- Report argues AI harms are structural and calls for regulatory intervention, antitrust action, and community organising.
- Primarily a US-oriented advocacy document; APS relevance is indirect but useful for understanding critical-AI discourse.
A researcher argues that multimodal scaling cannot achieve human-level AGI, citing limits in embodied cognition.
Key points
- The piece challenges assumptions underlying some AI capability forecasts - relevant to how agencies assess AGI risk timelines.
- Primarily an academic-conceptual argument; limited direct operational relevance for APS practitioners right now.
Week of 12 May 2025
MIT AI Risk Repository spotlights a 2022 Google DeepMind taxonomy of LLM risks across six domains and 20 subdomains.
Key points
- The taxonomy covers discrimination, information hazards, misinformation, malicious use, HCI harms, and socioeconomic harms - directly relevant to APS AI risk assessment work.
- The underlying paper is from 2022; the MIT blog post is a summary spotlight, not new research.
Week of 5 May 2025
Good Ancestors' May 2025 newsletter covers three distinct developments: US AI safety rollbacks, OpenAI's Stargate global expansion, and Federal Court GenAI consultations.
Key points
- The Federal Court of Australia item is directly relevant to APS practitioners - submissions were open until 13 June 2025.
- US deregulation under Trump is framed as increasing pressure on Australia and other middle powers to fill the governance gap.
Week of 21 April 2025
MIT's AI Risk Repository updated to 1,612 unique risk entries across 65 frameworks, now including multi-agent risks.
Key points
- The repository provides causal and domain taxonomies designed to support policy, auditing, and governance processes.
- A credible reference resource for APS agencies developing AI risk frameworks or audit criteria - freely accessible.