HomeAI · summaries
Storia in 2 fonti
New attack on ChatGPT research agent pilfers secrets from Gmail inboxes
Unlike most prompt injections, ShadowLeak executes on OpenAI's cloud-based infrastructure.
Confronto fonti
2 prospettive sulla stessa storiaTimeline cronologica
- ·
arstechnica.com
New attack on ChatGPT research agent pilfers secrets from Gmail inboxes
Unlike most prompt injections, ShadowLeak executes on OpenAI's cloud-based infrastructure.
- ·
theverge.com
Researchers turned ChatGPT rogue and it robbed secrets from Gmail
Victims didn’t have to click anything and were not even aware they’d been hacked.