Description of proposed project

Good Ancestors advocates for Australian-specific policies aimed at solving this century's most challenging problems. Currently, we focus about 80% of our efforts on AI safety. Our theory of change involves identifying opportunities for significant impact, a "policy window" (conditions that make change possible), and an angle for us to contribute.

When we identify an opportunity, we take a coordinated approach to create the best chance of change:

(1) We review research and proposals and adapt them to the Australian context. In the last 12 months, we've also begun doing our own research, both for direct impact and to boost credibility. The UK AISI funded our work on novel AI risks, and we collaborated with an Australian university to complete a comparative analysis of global catastrophic risks vs. traditional hazards (links below).

(2) We pursue "outside game" with like-minded organisations, communities, and the media. In the last 12 months, we've expanded our "outside game" to include church groups, animal advocates, the union movement, and youth advocates. Our CEO, Greg Sadler, has given oral testimony to the Senate, appeared on TV and radio to talk about AI safety and has regularly written for an Australian national security publication.

(3) We pursue "inside game" by directly engaging policy advisers and decision-makers. Over the last 12 months, our inside game work has gained more traction, with roughly 10x the number of meetings with elected officials, including more senior ministers and several deeper relationships. We see "echoes" of our outside game work via our inside game, with officials commenting that they'd heard "similar" ideas from other sources.

We have evidence that this model works, but can't share it all in public:

(1) We're actively working with Departments, staffers and MPs on specific issues.

(2) The government has used our briefs to inform internal decision-making.

(3) We've persuaded the Government to change rules around the operation of a fund so that grants can tackle global catastrophic risks.

(4) We've secured specific recommendations in Government reviews, including on issues like expanding tax-deductible donations to animal welfare and catastrophic risk reduction.

(5) Our work is cited by the Government, mainstream think tanks and industry groups.

(6) We met with and briefed Australian delegates to international forums.

(7) We've coordinated public health leaders to make calls for action relating to pandemic prevention, lab safety, engineered pandemics and funding for research into emerging technologies to improve indoor air quality.

Why are you qualified to work on this?

Our CEO, Greg, has had a 15-year career in the Australian public service, including working in senior positions relating to catastrophic risk reduction, the COVID-19 response, cyber and critical infrastructure security, counter-terrorism (including CBRN risks) and as a senior political adviser. Greg is also on the boards or advisory boards of several charities, including Global Shield, Ashgro, ALLFED, and Effective Altruism Australia.

The organisation is supported by an experienced team, including Nathan Sherburn, Luke Freeman (former CEO of Giving What We Can) and Emily Grundy, who have executive experience and have worked in fast-moving startup environments as well as the nonprofit sector and academia.

What would you do if not funded?

Funding this kind of work in Australia can be hard due to our charity laws and the lack of major philanthropy, with most international funders being US-centric (or sometimes UK and EU). Without sufficient public funding, we're likely to spend more key staff time on fundraising, which is particularly hard for a small organisation at a time when we have relatively short policy windows given AI timelines and the pace of policy change.

With support, our work will continue, but our capacity will be reduced.

How much money do you need?

Recent grants have allowed us to recruit Luke Freeman (former CEO of Giving What We Can) and Emily Grundy (former MIT AI governance researcher). We don't want more than half our funding to come from a single donor, meaning that, in practice, every dollar from platforms like this "unlocks" additional dollars from larger funders.

Supporting documents

We have published:

• Legal Zero-Days: A Novel Risk Vector for Advanced AI Systems

• Mapping Australia's Risk Landscape: A Comparative Analysis of Global Catastrophic Risks and Traditional Hazards

• Australian AI Legislation Stress Test: Expert Survey