In the intricate world of AI governance, where every decision has far-reaching implications, a curious case has emerged in Australia. A charity, Good Ancestors, is at the center of a debate over its funding sources and influence on AI copyright rules. This organization, with its seemingly altruistic mission, has raised eyebrows due to its deep connections with influential donors and its potential impact on the country's AI landscape. As the story unfolds, it's a tale of power, influence, and the complex interplay between philanthropy, technology, and policy.
The Charity's Mission and Its Impact
Good Ancestors, founded in 2022, claims to be dedicated to helping Australia's leaders act in the interests of future generations. With a focus on AI safety and pandemic prevention, the charity has become a prominent voice in Australia's AI policy debate. Its influence is evident in its submissions to federal and state inquiries and its seat on the Attorney-General's Copyright and Artificial Intelligence Reference Group. This group is tasked with determining how copyright should apply to AI model training, a decision that could shape the future of the industry.
The Funding Puzzle
The funding lineage of Good Ancestors is where the story becomes intriguing. In 2024, the charity received $243,000 from the Survival and Flourishing Fund, primarily backed by Skype co-founder Jaan Tallinn. Tallinn's connection to Anthropic, an AI giant, is significant. He led Anthropic's $125 million Series A funding round in 2021 and remains an investor. This funding, along with a subsequent $580,236 from Open Philanthropy (now Coefficient Giving), raises questions about the charity's true interests. Coefficient Giving was co-founded by Holden Karnofsky, who joined Anthropic in 2025 and is married to Anthropic president and co-founder Daniela Amodei.
The AI Training Permit Proposal
Good Ancestors' influence is further highlighted by its proposal for an 'AI Training Permit'. This permit would allow companies to train AI models in Australia with certainty that they would not be subject to copyright restrictions. While this might seem like a win for AI companies, it raises concerns. The permit holders would pay a tiered fee into consolidated revenue, respect an opt-out mechanism for Australian rights holders, and comply with the National Data Centre Expectations. However, the proposal has been criticized by Claire Pullen, chief executive of the Australian Writers' Guild, who argues that it benefits big tech rather than Australian creators.
The Lobbying Debate
The use of philanthropic networks to influence regulation is a tactic that has raised eyebrows. Nicholas Pickard, executive director of public affairs at APRA AMCOS, a music rights body, describes it as a long-established lobbying tactic. Good Ancestors' chief executive, Greg Sadler, rejects any suggestion that the charity is acting in the interests of AI companies, emphasizing that it receives no funding from them. However, the timing of the grants and the charity's influence on AI policy have sparked debates about transparency and the true motives behind its actions.
The Broader Implications
This case raises deeper questions about the role of philanthropy in shaping policy. It also highlights the complex relationship between technology companies and the organizations that advocate for them. The influence of donors with ties to AI giants on the copyright debate is a powerful reminder of the interconnectedness of these industries. As Australia navigates the AI landscape, it must balance innovation with the protection of creators' rights, a delicate task that requires careful consideration and transparency.
In my opinion, this story is a fascinating insight into the world of AI governance and the power dynamics at play. It raises important questions about the role of philanthropy in shaping policy and the potential conflicts of interest that can arise. As AI continues to evolve, these debates will only become more crucial, and it is essential to approach them with a critical eye and a commitment to transparency.
