BONUS: The super-hacker AI was released... then pulled

In April, a company called Anthropic claimed they had a new AI model that was too powerful to release. They called it ‘Mythos’, and said it could find new ways to hack computers which humans had overlooked for decades. Mythos, they said, was too dangerous for public consumption. Instead, Anthropic released Mythos to trusted partners in government and businesses, including in Australia.

Then, in mid-June, a public version of Mythos was made available. Anthropic said this version was safe for anyone to use; the U.S. government strongly disagreed. After only a few days the U.S. government banned foreigners from accessing the new AI model. Australians were cut off, and we don’t know when access will be resumed.

Today, we’re bringing back an interview from April with cybersecurity expert Dr. Suelette Dreyfus. What dangers does Mythos pose, and are we ready for a tsunami of AI-enabled hacking?

Follow The Briefing:

TikTok: @thebriefingpod

Instagram: @thebriefingpodcast

YouTube: @TheBriefingPodcast

See omnystudio.com/listener for privacy information.