Very excited about the work we contributed to the redesign of NSW’s AI Assessment Framework, released last week. It includes several world-leading design elements that reflect a more mature way of thinking about AI risk. Most AI risk frameworks don’t fail because they miss risks. They fail because they never narrow them in a principled…
As Australia’s representative on the international expert advisory panel, I’m glad to see the 2026 AI Safety Science Report released under the leadership of Yoshua Bengio. The findings will be discussed further at the India AI Impact Summit later this month. I will be there. Beyond what is written in the report, CSIRO’s Data61 continues…
In case you missed our year-end summary of the progress we’ve made in responsible AI at CSIRO’s Data61 in 2025, you can find it here. From launching the new Engineering AI Systems book, our practical guide to AI design and the operation of high-quality AI systems, to partnering with the Audit Office of NSW to…
Happy new year! The shutdown period is often when the noise drops and the big questions surface. This year, that quiet coincided with genuinely striking progress in AI coding agents, especially in the six to eight weeks following the release of Claude Opus 4.5, and its integration into Claude Code. We have seen world-class software…
This is a slightly delayed post because I was on annual leave when the National AI Centre released 𝗕𝗲𝗶𝗻𝗴 𝗰𝗹𝗲𝗮𝗿 𝗮𝗯𝗼𝘂𝘁 𝗔𝗜 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗲𝗱 𝗰𝗼𝗻𝘁𝗲𝗻𝘁 – 𝗔 𝗴𝘂𝗶𝗱𝗲 𝗳𝗼𝗿 𝗯𝘂𝘀𝗶𝗻𝗲𝘀𝘀. https://lnkd.in/gpBUmeCX Coming back, I saw the level of uptake, discussion, and genuine relief from the community, and I want to say how pleased I am with…
How can generative and agentic AI improve public auditing without weakening trust? CSIRO’s Data61 has recently entered a multi year partnership with the Audit Office of New South Wales to explore how AI can support government audit functions in a transparent, accountable, and responsible way. I wanted to rephrase and share this after seeing the…
As part of the international expert advisory group representing Australia, I’m pleased to see the second Key Update of the AI Safety Science Report, led by Yoshua Bengio, released (link in the comment). The official update captures the broad progress: more robust adversarial training, better AI-generated content tracking, wider adoption of frontier safety frameworks, and…
Most creators say they want protection from AI, yet a recent UK survey found over 90% would not opt out of AI access to their work. Why? Today’s visibility depends on online discovery and search, and those systems are powered by AI. If you opt out, your work may be safe from AI training, but…
I often share missed or surprising observations in my talks. Here are five of them, summarised as food for thought. 𝗠𝘆𝘁𝗵 𝟭: 𝗢𝘃𝗲𝗿𝘀𝗲𝗮𝘀-𝘁𝗿𝗮𝗶𝗻𝗲𝗱 𝗔𝗜 𝘄𝗼𝗻’𝘁 𝗮𝗹𝗶𝗴𝗻 𝘄𝗶𝘁𝗵 𝗔𝘂𝘀𝘁𝗿𝗮𝗹𝗶𝗮𝗻 𝘃𝗮𝗹𝘂𝗲𝘀.It’s intuitive to think globally trained models can’t reflect our norms. Yet when frontier models answer the same cultural questions posed to national cohorts, they align most…
Every confidential message you’ve sent may already be stored — waiting to be unlocked. This is the essence of the “𝗵𝗮𝗿𝘃𝗲𝘀𝘁-𝗻𝗼𝘄, 𝗱𝗲𝗰𝗿𝘆𝗽𝘁-𝗹𝗮𝘁𝗲𝗿” problem: data intercepted today could be decrypted once quantum computers mature. This week CSIRO’s Data61 released our new report, 𝙌𝙪𝙖𝙣𝙩𝙪𝙢 𝙎𝙖𝙛𝙚 𝙏𝙧𝙖𝙣𝙨𝙞𝙩𝙞𝙤𝙣: 𝙍𝙚𝙖𝙡𝙞𝙩𝙮, 𝙃𝙪𝙧𝙙𝙡𝙚𝙨 𝙖𝙣𝙙 𝙋𝙖𝙩𝙝𝙬𝙖𝙮𝙨, mapping the scale of the challenge and…
Director/Head of CSIRO’s Data61
Conjoint Professor, CSE UNSW
For other roles, see LinkedIn & Professional activities.
If you’d like to invite me to give a talk, please see here & email liming.zhu@data61.csiro.au