October 9, 2026, (Inside AI) — Two senior OpenAI executives appeared before Australia's Joint Select Committee on Artificial Intelligence in Sydney on October 6, marking a rare public confrontation between a leading AI developer and lawmakers pressing for binding rules. Adam Cohen, OpenAI's chief of economic policy, and Chief Strategy Officer Jason Kwon faced questions about safety, competition, and the pace of deployment. The hearing, held at Parliament House, underscored how AI governance has shifted from abstract debate to legislative scrutiny.
The session came as several governments advance mandatory safety testing and transparency requirements for frontier models. Australia's inquiry is examining economic and national security impacts, with a report expected later this year. OpenAI's testimony signals the company is willing to engage regulators directly, even as it resists prescriptive rules that could slow releases.
Cohen told the committee that the company supports "risk-based, proportionate regulation" and warned against a patchwork of national laws. He said OpenAI has expanded its safety evaluations and red-teaming before launching new models. Kwon added that the firm is working with governments on standards for model transparency and incident reporting.
The hearing followed OpenAI's recent release of a policy paper urging international coordination on AI safety. That document proposed shared testing protocols and a global registry of advanced systems. Australian lawmakers questioned whether voluntary commitments are sufficient, citing rapid capability gains and competitive pressure among labs in the United States, China, and Europe.
Committee members also raised concerns about market concentration. They asked whether a handful of firms control the data and compute needed to build frontier models. Cohen acknowledged the concentration but argued that open-weight models and cloud competition are expanding access. He noted that OpenAI's API is used by thousands of startups, which he said lowers barriers to entry.
The appearance drew attention because OpenAI rarely sends policy leaders to foreign legislative hearings. The company has faced mounting scrutiny since its 2023 governance crisis and subsequent restructuring. Its testimony in Sydney reflects a broader strategy to shape rules before they are written, rather than react after passage.
Australia's inquiry is one of several global efforts. The European Union's AI Act is being implemented in phases, with obligations for general-purpose models taking effect in 2025 and 2026. The United States has relied on executive orders and agency guidance, while China has issued interim measures for generative AI services. The lack of a unified approach has created compliance burdens for companies operating across borders.
Safety advocates argue that voluntary pledges have not prevented harmful uses, such as disinformation and cyberattacks. They point to incidents where models were jailbroken to produce dangerous instructions. OpenAI has responded by updating its usage policies and building detection tools, but critics say enforcement remains uneven.
Industry analysts note that OpenAI's engagement with Australia may also serve commercial interests. The company has expanded partnerships in the Asia-Pacific region, including with Japanese firms and Australian enterprises. Regulatory goodwill can ease market entry and reduce legal risk.
The hearing also touched on AI's impact on jobs. Cohen said the technology will displace some roles but create new ones, urging governments to invest in retraining. Kwon emphasized that AI can boost productivity in sectors like healthcare and education. Lawmakers pressed for specifics on transition support, and both executives said OpenAI would share data with policymakers.
OpenAI's appearance in Sydney did not resolve the central tension: how to balance innovation with safeguards. The committee is expected to recommend a mix of mandatory and voluntary measures. Whatever it decides, the hearing showed that AI companies can no longer avoid the public stage. They must defend their choices under oath, in front of cameras, and in real time.
As the global AI race continues, the question is whether speed and safety can coexist. OpenAI's testimony suggests the company believes they can, but lawmakers remain skeptical. The next round of hearings, expected in November, will test that belief further.
