Project summary

Organisation Details

Organisation - Sydney AI Safety Space (SASS)

About - The Sydney AI Safety Space is a co-working hub in Sydney, Australia, offering free office space for people working in the field of AI safety.

Fiscal Sponsor - Good Ancestors Ltd, a non-US charity with 501c3 equivalence determination.

Directors - Hunter Jay, Zac Broeren

Project & Funding Details

Minimum Viable - US$20,000 - This would allow us to continue offering co-working until July 2027.

Target - US$86,000 - This would fund the office space, a part-time community manager, and up to twelve AUD$6000 microgrants to be distributed to new AI safety projects throughout the year.

Amounts in between would allow a smaller scale version of the microgrants program to be run.

Funding Documents

Target Budget FY27 (USD) - The budget for the next 12 months, with and without the new proposed microgrants program.

Expenditure FY26 (AUD) - Historical expenditure for the previous financial year, primarily on office space.

SASS Utilisation - The utilisation of the space since detailed tracking began Q2 2026. Long term members are publicly listed on the website.

Microgrants Program Details - Details on the criteria and application process of the proposed microgrant program.

What are this project's goals? How will you achieve them?

We are seeking funding to continue offering free community office space to people working in AI safety in Sydney. The space has been running for a little over a year now, and has grown alongside the AI safety community in Sydney, helped produce orgs such as Lyptus Research, and given researchers a place to work while in town for events like the AI Safety Forum.

Having an office filled with capable people working on important problems is motivating. It makes it easier to get feedback and throw around ideas, it makes it possible for research groups and projects to form with people who meet here. It grounds our ideas and prods us to do better, and is a hub for news and events. It also provides a network where people can get help, mentorship, funding, and so on, outside of the Bay.

In addition to the primary goal, we are interested in running a micro-grants program. Over the last year, we have noticed several opportunities where a small grant (currently intending AU$1000 [US$716] per week for six weeks) would have helped support people at the early stages of projects, and reduced the risk for them to begin working on AI safety.

Had we been running this micro-grant program in FY26, potential candidates would have included Slava Chalnev's Cycle-Consistent Activation Oracle work (which was later referenced by Anthropic), Sean Peter's career transition and early cyber work -- which later became Lyptus Research -- and Stephen Elliott's multiagent systems work.

I think SASS as a co-working space is in a unique position here -- we see new projects appear at the earliest stages, and can make assessments of quality based on co-working with the individuals requiring funding rather solely via traditional paper applications. A low-overhead, rapid, micro-grant system -- similar to BlueDot Impacts Rapid Grants program -- might be able to significantly increase the odds that someone jumps into work full time, or does so sooner than they otherwise would be able to.

I also suspect that being able to offer micro-grants might make it easier for people to find and engage with SASS. While we have been growing (up to 17 members in April 2026, compared with 10 in June 2025), a small grants program increases our legitimacy and quality of our marketing further, and might lead to more high quality people finding and working out of SASS.

How will this funding be used?

At the low end, the primary expense is for office space. We expect to pay approximately US$56,000 for our office space usage over the next 12 months (an average occupancy of 8 desks between over 20 members), with approximately $14,000 of that accounted for with member fees, and $33,000 from existing cash on hand. A buffer for events, fiscal sponsorship fees, insurance, and amenities is added on top for a minimum viable ask of $20,000 USD for twelve months of operation.

With additional funding, we will be able to hire a part time office manager from within the community to take over some of the voluntary work I have been doing, and make sure the space is clean, tidy, and welcoming all the time. This might cost about $15,000.

Finally, the remaining ask is for a seed budget for the micro-grants program proposal, seeking USD$51,840 for this purpose. This would allow twelve AUD$6000 grants to be provided throughout the coming year.

Who is on your team? What's your track record on similar projects?

The main two SASS volunteers are Hunter Jay and Zac Broeren, but the space as a whole hosts people who work at the FAR.AI, Harmony Intelligence, the Australian AISI, Lyptus Research, and TARA, as well as multiple MATS Scholars, some university students, and a few current and former Atlassian engineers looking to enter AI Safety. We have a list of frequent members on our website, and utilisation is tracked here, but I'd like to call out some notable things SASS has supported in the past year:

• The founding of Lyptus Research, which received attention recently for replicating the METR time-horizons chart in the domain of offensive cyber. Sean Peters, Jack Payne, Slava Chalnev, as well as myself as a board member (Hunter Jay), were all working out of SASS independently prior to Lyptus being founded, and I think SASS can be assigned some credit for the coming together of this specific group on this work.
• The Sydney edition of the Apart AI Control Hackathon, run by Stephen Elliott, and with several other SASS members taking part.
• The Sydney AI Safety Fellowship, ran by Chris Leong and Jack Payne, which ran out of the space with support provided for food and amenities.
• The ‘Technical Alignment Research Accelerator’ (TARA), which is being run by members Zac Broeren and Yanni Kyriacos.

We have also been a focal point for people exploring AI Safety, including hosting a Q&A event with Brayden McLean of Anthropic, and meeting with guests who reach out to find out about AI safety -- including people from the USyd ‘US Studies Centre’ Director of Strategic Technologies and several students and other industry people.

We also hosted many visiting researchers attending the AI Safety Forum in Sydney, support monthly AI safety meetups, and are a founding member of an international network of AI safety hubs, including Singapore (SASH), Cape Town, and Toronto (Trajectory Labs).

What are the most likely causes and outcomes if this project fails?

Two possible failure modes;

• If the Sydney AI Safety community shrinks, or disperses to other hubs (primarily SF), then the utilisation rate would drop, and the space would cease having a critical mass of high quality people. This is less likely that it was a year ago, as we have grown over the previous 12 months, but it remains a potential problem that we're conscious of.

• If we are too broad, or grow too fast, there's a risk that the culture of SASS will be destroyed. We have a range of views, but currently lean very technical, and are quite ASI- or AGI-pilled. There is an expansion rate of new people entering AI safety that we can support, and bringing in new people who are unfamiliar with these ideas is one of our goals! However, there is a risk that as AI goes mainstream, the quality and depth of the conversation could drop, and we could enter a spiral where more experienced and knowledgable people stop wanting to come in. This is also not a major risk today, but it is something we keep an eye on.

How much money have you raised in the last 12 months, and from where?

We previously received grants from Brayden McLean for $10000 around Jun 1, 2026, and from AI Safety Tactical Opportunities Fund (AISTOF) for $38000 around Jun 17, 2025. This application is in addition to these amounts. A previous iteration, then called the Sydney AI Safety Hub and ran by Yanni Kyriacos under AI Safety ANZ, also recieved $5167 via Manifund in early 2025.