Data61 drops world-class seL4 security team


Joseph Brookes
Senior Reporter

The world-leading Australian research team that developed the extremely hard to hack seL4 microkernel has been disbanded by the CSIRO, with staff to be moved to AI projects or sacked in a restructure of Data61.

Work will continue on the project through an independent seL4 Foundation established last year but those involved say it will be much harder to attract funding and world-leading talent will be lost.

The changes are part of a restructure of CSIRO’s digital arm Data61, that will see 70 positions lost. The agency said different positions are being created in line with CSIRO’s new strategy built around AI, but it will take two years for headcount to return to current levels.

CSIRO to shutter ON startup program
The CSIRO has pulled funding from a world-leading security project as part of a new strategic focus on AI.

InnovationAus understands CSIRO staff were told by management at a meeting on Wednesday that the seL4 group known as Trustworthy Systems no longer fits the agency’s strategy and would be dissolved.

The decision to axe seL4, which has seen its team reduced over the last year, was revealed on Twitter on Friday by UNSW Scientia Professor Dr Gernot Heiser.

Dr Heiser helped develop seL4 and remains involved in the work through a collaborative research project between CSIRO and UNSW.

“Here is an absolutely recognised world class, world leading asset that’s unique in its composition and its track record and ability to do outstanding research, that’s being abandoned and destroyed,” he told InnovationAus.

The seL4 microkernel allows ironclad separation between software systems so that hackers cannot access the critical parts of the operating system by entering through a poorly protected hatchway.

It claims to be the world’s most highly assured operating system kernel.

“We have changed the notion of the possible and really shown the world that you can form the proof correct real world systems,” Dr Heiser said.

“We’ve done things no one else could do, and no one else even thought possible. That sort of unique thing is going to be destroyed, for sure.”

A spokesperson for the CSIRO confirmed the agency will no longer support Trustworthy Systems and will focus on cybersecurity and emerging areas like “Trustworthy AI”.

“The Trustworthy Systems group is focused on the area of formal methods for design, implementation, and verification of software systems,” the spokesperson said.

“It is mature area of technology that CSIRO has invested in over a number of years and is now well supported outside the organisation.”

The CSIRO has shifted its focus to artificial intelligence, with $124 million in new government funding for the technology announced in last week’s budget.

“They decided we don’t fit that [AI strategy],” Dr Heiser said.

“Of course we’re not AI. If you decide AI is all you want to do, then we are surplus to requirements.”

The science agency’s AI focus is “mind boggling strange”, Dr Heiser said, considering the exodus of leading Australian AI researchers, with many leaving for a far more lucrative private sector led by Big Tech.

He fears a similar scenario of leading Australian security researchers being pushed out of the local research sector.

Last year, the Trusted Systems group established a seL4 Foundation under the Linux Foundation to continue development and attract industry funding. The foundation already has active members and some industry support but CSIRO’s decision to abandon the project is a huge setback.

Australian National University Associate Professor Vanessa Teague said the decision to defund the world leading security research groups was difficult to hear.

“I’m just glad that Gernot has had the foresight to set up an independent foundation,” Dr Teague told InnovationAus.

“The foundation means that seL4 can continue to make profound and useful contributions to cybersecurity, without being dependent on the vagaries of government funding.”

The CSIRO said under its new strategy, investment is being directed at AI, digital science and technology as part of Australia’s COVID recovery, and “reinventing the way science is done using digital to revolutionise the future of scientific discovery”.

As a result of the changes the agency expects to create around 100 positions including 30 post doctorate positions, but several Data61 workers will lose their current positions.

“In the short term up to 70 people in Data61, including indefinite staff and term ends, will be potentially impacted, however the number will likely be less as we work to redeploy people throughout the organisation,” a spokesperson for the CSIRO said.

“Within two years, given the new positions, we expect headcount to be at similar levels to today.”

Do you know more? Contact James Riley via Email.

6 Comments
  1. 6R1M0R4CL3 4 years ago

    They took the right decision. Seriously, no one is using that so-called unhackable kernel you speak of. And just wrting “unhackable kernel” makes us hackers laugh to tears. You have no grasp of what the world runs on (mostly, Linux) and sel4 is a research toy no one uses nor cares about in the world of security or hacking.

    • m%D$8Ai2 4 years ago

      I have no love for Data61 – but this comment didn’t really add much here…

      “sel4 as a ‘research toy” – CSIRO is a research organisation. Research toys are part of what they do – muck around with something new and see if it makes sense. Strip what’s useful and dump the rest. That’s the natrue of tech progress.

      And ‘us hackers’? I… just can’t even…

  2. Daniel 4 years ago

    This is very sad to hear. I would say that the seL4 project is an inspiring project demonstrating how formal methods and formal proofs are able to provide unique guarantees that greatly increase the trust we can have in the security of systems built upon the seL4 microkernel.

    Hindering future progress on this and related ideas seems like an unnecessary consequence of putting greater focus on AI. I agree that AI is important, but is it really the only important research area?

  3. Johnno 4 years ago

    Classic CSIRO management – build up critical mass in science and then throw it away when the funding changes – I recall the semantics and reasoning (AI) group was dismantled at the last restructure and here they are rebuilding again!

  4. Annie 4 years ago

    Really shocking news! There are lots of labs dismantled at the same time including the WIFI invented lab. Investment in AI seems a shortsighted and misguided decision.

  5. Evan Rowley 4 years ago

    This is shocking news. I am saddened that development of the seL4 microkernel will cease. Investment in AI seems a shortsighted and misguided decision.

Leave a Comment