OpenAI chief executive Sam Altman was “not aware” of a breach by one of its models of a Medicare data portal when he met with Deputy Prime Minister Richard Marles, a parliamentary inquiry has been told.OpenAI chief strategy officer Jason Kwon flew from the United States to appear before the Joint Select Committee on Artificial Intelligence in Sydney on Tuesday.It comes just weeks after it was revealed an OpenAI model accessed statistical Medicare data while in training, prompting concerns from local and international leaders.Speaking outside NSW parliament on Tuesday, Mr Kwon said he was going to answer questions for the committee “and set forth how we are going to proceed”.“I’d like to reiterate that we’re sorry for what’s happened,” he said.“We’re here to talk about how we are going to do better for Australia.Others you may likenewsnews“It’s important time for AI capabilities, and it’s on us to improve what we do. So we look forward to working with the committee today”.OpenAI chief strategy officer Jason Kwon will front the parliamentary inquiry. Picture: NewsWire / Nikki Short. Picture: NewsWire / Nikki ShortOpenAI chief executive Sam Altman was unaware of the AI breach of a Medicare portal when he met Deputy Prime Minister Richard Marles. Picture: Alexi J Rosenfeld / Getty Images via AFPAppearing before the committee later, Mr Kwon said Mr Altman was “not aware at the time” of the breach during a September 1 meeting in Silicon Valley with Mr Marles. “I agree that the process by which people became aware of this incident inside our company could have been much better,” he said.“And we want to make sure that something like that does not happen again.”Asked by senator David Pocock why someone at OpenAI did not reach out to a minister in government, rather than through a generic email, Mr Kwon said: “I think that in retrospect, we should have done what you were suggesting”. “So the reason why it happened the way that it did is I think people were thinking about this as a technical situation, and they wanted to contact the technical counterparties.“But it’s not good enough.”‘Slow down”Asked by Liberal frontbencher Andrew Hastie about calls from within the sector to slow down development of frontier models, Mr Kwon said AI capabilities were “going to grow”.“The safety and alignment that needs to go with the capabilities needs to keep pace, and that’s what we mean by pacing the frontier,” he said, noting Mr Altman’s support for “pacing” frontier development.Mr Kwon was also grilled on whether legislators were keeping pace with advancements in the technology.“With the tension in the space, legislators are increasing their attention … and I think that’s generally a good thing,” he said.“But I think that what also needs to happen is that the representative of society needs to make more decisions for, you know, so that we are not making all these decisions. “And I think one of the key things here is that the technology is powerful and we want to ensure that lawmakers and policymakers and democratic societies have a say in how this technology develops.”OpenAI exec speaks outPressed on whether there was an example leaders should look to in terms of working with AI companies, Mr Kwon noted co-operation between the US government and OpenAI in which there is “access by the government to understand the model capabilities early on”.“And we provide the government of the latest capabilities so that they can start to use it sooner than anyone else.”OpenAI national security adviser Peter Antsee said Australia had been “an exemplar in terms of outreach and engagement, both from government and from private sector”Earlier, Mr Kwoon said it was possible the breach may never have been discovered if OpenAI had not alerted the government.“It’s possible that somebody might have discovered some activity online, but not necessarily been able to attribute it to our agents,” he said.Mr Kwon said OpenAI’s response had focused on “strengthening research, security, monitoring and model alignment, as well as improving how we identify, escalate and respond to incidents”. OpenAI had also committed resources and technical assistance to affected agencies to help strengthen cyber defence.“We’re also establishing a local task force here in Australia with independent Australian expertise to make recommendations on how to better manage the risks associated with increasingly capable AI,” he said.Mr Kwon said additional precautions and safeguards had not been “on the radar of our company as well as a lot of others” prior to the incident.“Clearly it is now,” he said.The AI giant has since put on pause its latest model.“ … because of the current situation, we wanted to make sure we were putting our best foot forward,” he said.“So we decided to not release that model and instead go back to training.”OpenAI gave evidence to member of the Joint Select Committee on Artificial Intelligence at NSW Parliament House on Tuesday. Picture: NewsWire / Nikki ShortWhile Mr Kwon admitted OpenAI should have handled its communication with the government better, he said OpenAI “wanted to understand more of the facts before we spoke to the impacted parties”.He noted after an incident involving the NSW Parks and Wildlife Service, the government branch was instead notified within 48 hours. As part of its review following the Hugging Face incident – an earlier breach by an OpenAI model – OpenAI sought to ensure it had “a better view of all the activity that may have occurred in relation to agents accessing the internet”. That would include using models to access other model’s activity online. “It’s about 50PB (petabytes) of data, which if you convert into a person going through the information, it would take that person about 66 million years, never sleeping, never eating, never taking a break to get through all the information,” Mr Kwon said.“We are using our own AI models and systems to be able to go do this backwards looking work.”Mr Kwong also boasted of OpenAI’s deepening ties with Australian businesses and researchers, including with supermarket giant Coles.“Roughly half of the Australian adults use ChatGPT every week, while Australia is also leading the way globally in business adoption of AI,” he said.“We want Australians to have the skills to make the most of AI and adapt as the technology evolves.”In an earlier blog post, OpenAI said Australia’s governments, industries and citizens “are and have been invaluable partners to OpenAI”.“We do not take this for granted, and we intend to make this right,” it said.The June incident was announced by Anthony Albanese on September 24. Authorities were notified through a generic Services Australia email address used for reporting potential breaches on September 10, despite OpenAI knowing of the breach in August.Both officials and OpenAI have said no sensitive or personal information was breached.The AI agent, tasked with research, also attempted to access the NSW Bureau of Crime Statistics and Research (BOCSAR), Victorian Department of Health and the Australian Institute of Health and Welfare.Other AI leaders Anthropic, Microsoft and Google will also appear before Tuesday’s hearing.Voice actor blames AI for losing workA voice actor has told a parliamentary inquiry she has been forced to take a second job after her acting work diminished amid the rise of artificial intelligence.MEAA Equity NSW president Jacqui Duncan told inquiry in Sydney on Tuesday that her voice acting work had declined over the past 18 months, and she did not know whether her voice had been used to train AI systems.Australian Recording Industry Association chief executive Annabelle Heard and Australian Association of Voice Actors president Simon Kennedy warned of AI’s effects on artists. Picture: NewsWire / Nikki ShortMs Duncan said she had never consented to her work being taken for AI training and wanted greater control over how it was used.She rejected suggestions AI could remove the need for human creators, arguing AI models relied on the experience and performances of people whose work had been used to train them.MEAA research and policy lead Lilia Anderson said workers often did not own the copyright to their work, meaning licensing agreements between rights holders and AI companies would not necessarily result in payments flowing back to the journalists, musicians, actors and other creators who produced it.The committee was also told there was no known remuneration being paid in Australia for the training of frontier AI models, prompting Labor MP Tim Watts to question whether the status quo was working for either creators or the development of an Australian AI industry.Screenrights chief executive James Dickinson said the existing copyright system worked better for larger organisations than smaller rights holders, which could struggle to negotiate with AI companies, while arguing the current system could be improved.Artists ‘roadkill’Australian artists could become “roadkill” in the rush to embrace artificial intelligence unless creators retain control over their work, the music industry has warned.Senator David Pocock gets his point across. Picture: NewsWire / Nikki ShortARIA and PPCA chief executive Annabelle Herd told the inquiry that creators risked being offered token payments for their work while those unwilling to participate could be left behind.“In other words, Australia’s artists will be the roadkill in the rush to this AI deal,” Ms Herd said.Australian Association of Voice Actors president Simon Kennedy said members had reported a 62 per cent decline in work over the past 12 months, arguing self-regulation was failing to protect performers as AI-generated commercials replaced jobs.Australian Writers’ Guild group chief executive Claire Pullen accused large language models of being built using creators’ work without permission and called for creators to have both the right to refuse its use in AI training and the right to be paid when they consented.The creative groups also pushed back against suggestions a levy on AI companies could adequately compensate workers, with Ms Herd arguing distributing $300m across affected creators would leave individuals with a “minuscule” amount.Mr Kennedy called for greater transparency around AI-generated material and said Australia should enforce protections for creators rather than weaken them to accommodate technology companies, while Ms Herd argued AI companies operating in Australia should be subject to Australian regulation regardless of where their models were trained.Call for AI companies to pay upAustralian media and entertainment workers have called for AI companies to pay a targeted fee to compensate creative workers whose work is used by the technology.Media, Entertainment and Arts Alliance acting chief executive Adam Portelli told the inquiry that Australia’s approach to AI needed to recognise the value of human creative labour and give workers the right to control and be compensated for the use of their work.He said musicians and voice actors were already having their work and likenesses replaced by AI, reducing employment opportunities in the creative industries.The union proposed an “equitable remuneration scheme” funded through a targeted fee on AI companies, with revenue used for direct payments to creative and media workers and a cultural fund supporting Australian entertainment and journalism.ABC senior policy and regulatory affairs executive David Sutton said human oversight was still vital for published material. Picture: NewsWire / Nikki Short“We (MEAA) are not arguing against technology,” Mr Portelli said. “MEAA is arguing for a system in which technological development does not come at the expense of the people whose creativity, skill and labour helped make the technology possible and whose work is essential for Australia’s cultural fabric.”Screenrights chief executive James Dickinson separately said about two-thirds of members surveyed were interested in licensing their work for AI use, but almost 90 per cent of those said they did not have the ability to negotiate those licences themselves.He said smaller rights holders needed assistance to counter the overwhelming “market power” of major technology companies, while Screen Producers Australia told the inquiry transparency and meaningful consent would be critical to ensuring creators received reasonable compensation. ‘Moving carefully’: ABC’s AI vowThe ABC has told a parliamentary inquiry it has no plans to replace journalists with artificial intelligence as it takes a cautious approach to adopting the technology.ABC senior policy and regulatory affairs executive David Sutton told a Joint Select Committee on Artificial Intelligence in Sydney on Tuesday that anything published by the national broadcaster continued to require human oversight.A rogue OpenAI agent breached a Medicare statistics portal. Picture: NewsWire / Nikki Short“Anything that goes out publicly has to be overseen by a human,” Dr Sutton said.“We’re taking a path of moving slowly and carefully with the technology so that we engage with it, but we are not rushing for it or overengaging.”The ABC uses Anthropic’s Claude as a productivity tool and has developed an internal AI tool to help staff find issues and material previously covered by the broadcaster.The broadcaster also urged parliament not to introduce an “opt-out” copyright system for AI companies, arguing existing copyright laws allowed developers to negotiate licences with rights holders and an opt-out model would unfairly place the burden and ongoing surveillance costs on content creators.The ABC also warned AI-generated search summaries were contributing to declining traffic to its content, although Dr Sutton said the broadcaster did not yet have firm figures on the scale of the decline.He said users could be satisfied with an AI-generated description without clicking through to the original ABC material, while the broadcaster warned inaccurate summaries could also pose a reputational risk.But Dr Sutton said AI could improve journalism by helping reporters conduct research and analyse large sets of documents, making those tasks more manageable without replacing their work.ABC head of content legal Kate Gilchrist rejected suggestions an opt-out copyright regime would solve the issue for creators, saying ABC content often incorporated material belonging to multiple rights holders rather than being wholly owned by the broadcaster.She said this meant the ABC could not simply scour the internet to ensure its material had been opted out across every place it was published, strengthening the broadcaster’s argument for AI companies to instead negotiate licences with rights holders.‘Big questions’Committee chair Jo Briskey said the hearing would give parliamentarians an opportunity to demand answers but also get clarity on “what happens next”.“It’s important we ask the questions not just about what happened but more importantly what happens next – how do we make sure that this incident does not happen again,” Ms Briskey told ABC News Breakfast.“It’s an exciting time, but artificial intelligence has its risks as well as its opportunities so it will be critical that the people that come before us today are ready to answer the questions that we have for them.”Assistant Minister for Science, Technology and the Digital Economy Andrew Charlton touted the inquiry as a display of Canberra putting politics aside to work in the national interest.OpenAI has is one of the world’s leading artificial intelligence firms. Picture: AFP / Pau BarrenaHe said “big tech has some big questions to answer”.More Coverage‘Critical’: Big call after Medicare AI breachNathan SchmidtAussies ‘rightly worried’ about AIJoseph Olbrycht-Palmer“We have set this up as a government in a bipartisan way, to make sure that Australians get answers to some of these questions from the big tech companies,” he told Seven’s Sunrise.“The government is also implementing some very tough standards that will make sure that these companies are accountable to the Australian people.“That’s a principle that the Prime Minister laid out a few months ago, and those standards will be out before the end of the year.”