An AI broke out of its cage & hacked a company. The world's spies warned us

OpenAI admitted this week that two of its most powerful models broke out of a sealed test environment and hacked another company - on their own initiative, to steal the answers to an exam they were sitting.  The AI lab says it’s unprecedented but it lands about a month after the Five Eyes spy alliance - the US, UK, Canada, New Zealand and Australia - warned that advanced AI is accelerating the speed, scale and sophistication of cyber threats, and that the shift would take months, not years. Australia's own signals directorate put its name to that warning. So how worried should we be? And what can any of us actually do about it?   In this bonus episode of The Briefing, we revisit Chris Spyrou’s conversation with cybersecurity expert Alastair MacGibbon, the founding head of the Australian Cyber Security Centre, recorded in the days after the Five Eyes warning was issued. Follow The Briefing: TikTok: @thebriefingpodInstagram: @thebriefingpodcast YouTube: @TheBriefingPodcastSee omnystudio.com/listener for privacy information.