Kangaroo LLM promises an open source, you-beaut ridgy-didge, true blue, all-Aussie LLM - by scraping your .au websites

What's that Skip? The tech industry has identified the need for an Aussie-flavoured foundation language model because no government department or even CSIRO are working on one? Enter Kangaroo LLM, a consortium of big tech players that include HP, Katonic and RackCorp. The pitch is that Kangaroo LLM will be open source (Apache 2 licensed, according to the T&Cs), however I can't find a single line of code on GitHub for Kangaroo LLM as yet, leading me to be sceptical. Kangaroo LLM was recruiting for two specialised "volunteer" roles, including a Bot Manager and Data Engineering Manager - although those roles no longer appear on LinkedIn - likely due to the ACS calling this practice in to question in a recent post. Kangaroo LLM are also calling for "volunteer contributors" to help scrape .au website data "ethically". Personally, I think it's great that industry has identified the need for an Australian-based LLM / foundation model, however this approach definitely feels like open-washing to me. I am pleased though to see that the company has provided instructions on how to block their bot from scraping your websites. What's the business model? Will the data be open sourced? Who will have access to the LLM? And how do we increase representation of all Australians in foundation models? In my view, a truly open source Australian foundation model would be the product of industry, academia, government, and the open source community working together to align with pieces such as the Voluntary AI Safety Standard and CSIRO's Artificial Intelligence Roadmap. Will I be volunteering to help them at this stage? Yeah, nah.


If you liked this tiny snippet of content from The Sizzle - Australia's favourite daily email containing the latest tech news & bargains - then sign up for a 30-day free trial below. No credit card required! Learn more about The Sizzle at https://thesizzle.com.au