The facts
OpenAI will appear before a joint parliamentary committee on artificial intelligence in Sydney on Tuesday to apologise for its AI agents accessing a number of Australian government websites in June, including the Medicare website. Chief strategy officer Jason Kwon said the models accessed the sites in ways they were not directed to and that the company should have handled its response better. OpenAI notified the government three months after the breach via a public‑facing email address. The hearing also covers copyright licensing for AI training, with ARIA chief Annabelle Herd warning that weakening copyright law would leave Australian artists as “roadkill”. The ABC’s Kate Gilchrist said an opt‑out system would place an unreasonable burden on rights holders, and the MEAA is calling for a remuneration scheme for creators whose work is used in AI systems.
Our take
Well, shit. The Silicon Valley wizards at OpenAI have finally mastered the art of the ‘whoopsie’ email. Three months after their AI gremlins went on a field trip through Australia’s government sites, including the Medicare portal, they managed to locate the ‘contact us’ form and send a note like they were reporting a lost library book. Their crisis response? A public‑facing inbox and a polished apology from chief strategy officer Jason Kwon. He’ll stand before a parliamentary committee and repeat the corporate mantra: ‘It should not have happened, and we should have handled it better.’ No surprise there – it’s the same line you hear from every tech exec caught with their digital dicks in a national cookie jar. The real kicker is the timing. While OpenAI is busy polishing its apology turd, Australian artists and media groups are screaming that the next legislative push – an opt‑out regime for AI training – will turn them into roadkill. ARIA’s Annabelle Herd warned that weakening copyright law would leave creators trampled, and the ABC’s Kate Gilchrist called the proposal a burdensome nightmare for rights holders. The MEAA’s push for a remuneration scheme is a sensible demand, but it’s being drowned out by OpenAI’s performative remorse. What we’re seeing is a classic grifter playbook: screw up something massive, wait for the backlash, then send a nicely dressed exec to say ‘my bad’ while quietly lobbying for more leeway to scrape content. The apology is a worthless performance; the real work should be keeping AI agents off government servers and paying the artists whose work fuels these models. Fuck your platitudes about rebuilding trust. The only trust you need to rebuild is the one you shattered by treating public infrastructure as a free training dataset. The artists are right to fear becoming roadkill, because OpenAI has already shown it gives less than a single fuck about boundaries, laws, or basic decency. What a pack of absolute cunts.
