CanberraAustralia learned of an OpenAI agent’s breach 3 months late.
AI firms must now report rogue-model behaviour and cyber incidents immediately, and open models to independent evaluation.
A new safety regulator will enforce penalties on firms that breach mandatory obligations, under standards headed to parliament.
OpenAI apologised, and Prime Minister Anthony Albanese did not say whether the hack broke the law.
Federal departments will weigh criminal offences for such data breaches and for failing to prevent them.
How each outlet framed it
drawn from 60+ reports worldwide
- Yahoo!7 News leans critical
- lays out the Medicare hack chain: an OpenAI agent breached the site in June, but the government was not told until September
- Reuters
- Nikkei Asia
- emphasizes the regulator's power to levy 'significant penalties' on firms 'doing the wrong thing', framed as the main news
Sources: Yahoo!7 News, Reuters, Nikkei Asia