『Ep. 191: Tim Watts MP on AI sovereignty, safety and Australia’s choices』のカバーアート

Ep. 191: Tim Watts MP on AI sovereignty, safety and Australia’s choices

Ep. 191: Tim Watts MP on AI sovereignty, safety and Australia’s choices

無料で聴く

ポッドキャストの詳細を見る
In a somewhat different episode of the podcast, Darren speaks with Tim Watts MP, a member of the parliament’s new Joint Select Committee on Artificial Intelligence, about how Australia should think about AI at a moment when both the technology and the politics surrounding it are moving extraordinarily quickly. The conversation (recorded on 11 Sep) begins with the remarkable run of AI news over the past fortnight: the investigation into OpenAI agents coordinating to hack Hugging Face, increasingly stark warnings from researchers inside frontier AI companies about loss-of-control risks, evidence of attempts to use advanced models for dangerous biological research, and extraordinary new claims about AI capabilities in mathematics. Darren and Tim discuss how non-specialists—and governments—should think about risks that are deeply contested, difficult to quantify and potentially very large. They then turn to Australia. What does “sovereignty” actually mean in AI when Australia cannot realistically control the entire technology stack? Tim distinguishes between hosting compute in Australia, controlling infrastructure, and securing access to frontier models. He discusses the different value of frontier capability, cost and optionality; open-weight models; the distinction between training and inference; and why hosting significant training compute could give Australia a more consequential place in the emerging AI ecosystem. The discussion also explores the limits of Australian leverage. A relationship with an AI company is not the same thing as leverage over the government that regulates it. Darren and Tim draw parallels with the management of Australia’s alliance relationships and Allan Gyngell’s argument that Australian statecraft should seek to expand the country’s options and room for manoeuvre. They also consider what Australia should demand in return for hosting very large AI investments: access to compute, research partnerships, safety evaluation, domestic skills and capability, alongside the local questions of energy, water, infrastructure and social licence. Finally, they look beyond Australian sovereignty to what Australia might contribute internationally—through AI safety research, cybersecurity, cooperation among middle powers and potentially the Australia Group’s work on biological and chemical weapons. Tim closes with advice for people who are not technologists but want to understand AI: use the tools, learn by doing, and read some science fiction. Australia in the World is written, hosted and produced by Darren Lim, with research and audio editing by Hannah Nelson and theme music composed by Rory Stenning. Relevant links The parliamentary inquiry and Australian policy Joint Select Committee on Artificial Intelligence: https://www.aph.gov.au/Parliamentary_Business/Committees/Joint/Artificial_Intelligence Australian Government, National AI Plan: https://www.industry.gov.au/publications/national-ai-plan Australian Government, Expectations of data centres and AI infrastructure developers, including “Expectation 5” on research, innovation and local capability: https://www.industry.gov.au/publications/expectations-data-centres-and-ai-infrastructure-developers Prime Minister Anthony Albanese, “AI in Australia’s interests”, 15 July 2026: https://www.pm.gov.au/media/ai-australias-interests-0 Australian Government–Anthropic MOU on AI safety and research: https://www.anthropic.com/news/australia-MOU Australia–UK cooperation between their AI safety/security institutes: https://www.minister.industry.gov.au/t-ayres/media/australia-and-uk-partner-strengthen-ai-safety-and-security Deputy Prime Minister Richard Marles on AI training, investment and sovereignty following meetings with US frontier labs: https://www.minister.defence.gov.au/transcripts/2026-09-02/doorstop-washington-dc Pax Silica Declaration: https://www.industry.gov.au/publications/pax-silica-declaration-countries-attending-pax-silica-summit-12-december-2025 Penny Wong, Allan Gyngell Lecture, 14 August 2026—the source of the discussion about expanding Australia’s options and maximising its choices: https://www.foreignminister.gov.au/minister/penny-wong/speech/allan-gyngell-lecture Recent AI safety and capability developments METR, independent investigation of the OpenAI/Hugging Face incident: https://metr.org/blog/2026-08-26-openai-hugging-face-incident-investigation/ OpenAI’s account and post-mortem of the Hugging Face incident: https://openai.com/index/hugging-face-incident-and-the-road-ahead/ Anthropic, Countering misuse of AI: September 2026, including biological and cyber misuse cases, September 2026: https://www.anthropic.com/threat-intelligence-report-september-2026 Jacob Coxon on leaving Anthropic over AI safety concerns, 9 September: https://www.axios.com/2026/09/09/anthropic-researcher-ai-warning-interview New York Times, “A.I. Could Possibly End Humanity. How Are ...
adbl_web_anon_alc_button_suppression_t1
まだレビューはありません