Week of 9 February 2026
A randomised trial of 1,298 participants found LLMs performed no better than search engines for medical decision-making.
Key points
- LLM benchmark scores failed to predict real-world performance, raising questions about reliance on standardised evaluation methods.
- UK-based research with no immediate Australian regulatory parallel, though findings are relevant to health AI risk assessment globally.
Alan Turing Institute report warns the UK must act urgently on AI-driven information threats during crisis events.
Key points
- Focus is on AI-amplified misinformation and disinformation risks in high-stress, time-sensitive contexts like disasters or emergencies.
- Limited extracted text available; APS relevance depends on recommendations - worth monitoring rather than acting on.
MIT AI Risk Repository spotlights a 2023 safety taxonomy for Chinese LLMs covering 8 harm scenarios and 6 adversarial attack types.
Key points
- The taxonomy claims scalability beyond Chinese-language models, making it potentially relevant to broader LLM safety evaluation work.
- This is a blog summary of a 2023 academic paper - useful reference material, not new guidance or policy.
SafetyBench is a bilingual benchmark assessing LLM safety across 7 risk categories using 11,435 multiple-choice questions.
Key points
- The MIT AI Risk Repository spotlights this as one of 28 frameworks cataloguing AI risks - useful for comparative evaluation work.
- A 2023 academic paper; this blog post adds no new findings beyond summarising the original arXiv publication.
Google-affiliated researchers find LLM reasoning models implicitly simulate multi-agent 'societies of thought' when solving hard problems.
Key points
- ChipBench benchmark reveals frontier models still perform poorly at real-world chip design tasks, despite hype around AI-driven hardware.
- AI research newsletter content; limited direct APS governance or policy relevance, included for technical context.
NIST NCCoE has released draft SP 1800-39 on data classification practices, open for comment until 30 March 2026.
Key points
- The publication frames data classification as foundational to secure AI model training, Zero Trust, and quantum-safe cryptography.
- Limited direct relevance to APS AI governance work; primarily a US data-security standard with peripheral AI framing.
Week of 2 February 2026
NIST's NCCoE is consulting on a concept paper addressing identity, authorisation, and auditing of AI agents.
Key points
- The paper seeks input on use cases, standards, and controls including prompt injection mitigations for agentic AI.
- Public comment closes 2 April 2026; Australian agencies with agentic AI programs could contribute or observe.
MIT AI Risk Repository spotlights a Google DeepMind-led paper on ethical risks of advanced AI assistants.
Key points
- Framework covers value alignment, human-assistant interaction risks, and societal-scale impacts across three structured areas.
- Identifies an 'evaluation gap' where current approaches focus on model-level considerations rather than broader sociotechnical effects.
A 2023 paper proposes embedding model evaluation for dangerous capabilities and alignment into AI governance processes.
Key points
- Nine dangerous capability categories are identified, including cyber-offense, deception, self-proliferation, and situational awareness.
- MIT AI Risk Repository surfaces this as one of 25 risk frameworks - useful reference material for agencies building AI risk taxonomies.
Singapore's AI Verify Foundation developed an 11-principle testing framework covering transparency, safety, fairness, and accountability.
Key points
- The framework aligns with ASEAN, EU, OECD, and US AI governance frameworks, giving it cross-jurisdictional reference value.
- This item is a MIT AI Risk Repository blog spotlight - the substantive content originates from a 2023 Singapore Government document.
Import AI #443 is a multi-topic newsletter covering agent ecologies, AI R&D automation risks, productivity evidence, robotics, and brain emulation.
Key points
- The AI R&D automation section is the highest-signal item: a CSET workshop report warns of compounding strategic surprise and declining human oversight.
- Limited direct operational relevance to Australian federal agencies; most value is as a horizon-scanning signal across frontier AI themes.
Week of 26 January 2026
NIST CAISI has released draft NIST AI 800-2, proposing best practices for automated benchmark evaluations of language models.
Key points
- The draft targets AI deployers, developers, and third-party evaluators - including procurement specialists using evaluation reports.
- A 60-day public comment period closes 31 March 2026; Australian agencies or evaluators could submit feedback.
Alan Turing Institute research argues a thriving AI assurance marketplace is essential to UK defence AI adoption and economic growth.
Key points
- The UK defence AI assurance framing has parallels for Australian Defence and APS agencies developing AI risk frameworks.
- Extracted text is truncated; full research substance cannot be verified from available content.
Frontier AI models can now automate exploit generation for software vulnerabilities, signalling a shift toward machine-speed cyberoffence.
Key points
- A Stanford economist argues AI warrants existential-risk spending equivalent to 5–10% of GDP annually, including a compute tax.
- US labour research finds clerical and administrative workers face the worst AI displacement risk with the least capacity to adapt.
Week of 19 January 2026
The Alan Turing Institute has released a framework and self-assessment tool for UK AI regulators.
Key points
- The tool is designed to help regulators evaluate their own capacity to oversee AI effectively and responsibly.
- Limited extracted text constrains full analysis; the underlying source warrants direct review for detail.
Jack Clark's essay describes firsthand experience deploying AI research agents to automate large-scale literature analysis and task execution.
Key points
- Drexler's 'Framework for a Hypercapable World' argues good AI outcomes depend on building institutional structures, not controlling singular AI entities.
- Content is primarily analytical and reflective; limited direct APS applicability but carries useful framing for AI governance thinking.
A year-in-review podcast with Air Street Capital's Nathan Benaich covers 2025 AI progress, regulation, and investment trends.
Key points
- Topics include sovereign AI requirements, EU AI Act compliance gaps, export controls, and open-weight model safety — all relevant to Australian AI strategy context.
- This is a VC-investor perspective podcast; it offers useful framing but limited direct APS applicability.
Alan Turing Institute announces an AI-powered climate and food security forecasting initiative for West Africa.
Key points
- Initiative applies frontier AI to humanitarian and agricultural resilience outcomes - a use case with potential policy interest for CSIRO and DFAT.
- Item is thin on detail given truncated text; full substance requires reading the source directly.
NIST's NCCoE has released its inaugural Project Portfolio outlining active cybersecurity research priorities and projects.
Key points
- The portfolio covers US cybersecurity innovation broadly; AI-specific content is not confirmed as a primary focus.
- Limited direct relevance to Australian federal agencies - included for context as a US standards-body output.
Week of 12 January 2026
NIST's CAISI has issued an RFI on securing AI agent systems, with submissions closing 9 March 2026.
Key points
- The RFI targets risks unique to agentic AI: prompt injection, data poisoning, misaligned objectives, and specification gaming.
- Outputs will inform voluntary US guidelines - a likely reference point for Australian agentic AI governance work.
Import AI 440 covers four distinct research items: adversarial LLM evolution, AI-automated compliance, o-ring labour economics, and LLM persuasion of conspiracy beliefs.
Key points
- The automated compliance piece proposes 'automatability triggers' - regulations that activate only once AI can cheaply enforce them - directly relevant to AI governance design.
- The LLM persuasion research and labour economics item have indirect APS relevance; the adversarial evolution item is primarily technical interest.
NIST NCCoE hosted a hybrid workshop on 14 January 2026 to develop its Cyber AI Profile under the Cybersecurity Framework.
Key points
- The preliminary Cyber AI Profile and SP 800-53 COSAiS overlay are open for public comment until 30 January 2026.
- This is a past event with a closed comment window - limited immediate action available for APS readers.
AI Now Institute publishes a series critically analysing the 2026 India AI Impact Summit and its governance discourse.
Key points
- Contributors examine how terms like sovereignty, democratisation, and accountability are used in AI policy debates.
- Limited direct relevance to APS practitioners; useful background on critical perspectives in global AI governance framing.
Week of 5 January 2026
DTA's updated Policy for the responsible use of AI in government came into effect 15 December 2025 for all non-corporate Commonwealth entities.
Key points
- New mandatory requirements include internal AI use-case registers, accountable owners, AI impact assessments, and foundational AI training for all APS staff.
- First mandatory requirement begins 15 June 2026; all remaining requirements take effect December 2026, giving agencies a staged implementation window.
Good Ancestors' January 2026 newsletter covers Australia's AISI hiring, Grok deepfake crisis, Productivity Commission AI regulation findings, and global safety warnings.
Key points
- Multiple items directly affect APS work: MYEFO reveals $166M GovAI Chat, AISI founding team roles, ACCC agentic AI warnings, and automated welfare liability.
- Roundup format means each item warrants separate engagement at source; this is a curated signal, not a single-issue analysis.