OpenAI tightens model training monitoring after Medicare data breach
Quoting Victoria Kim
AI BriefingOpenAI has added monitoring measures following the Medicare data breach, allowing employees to “intervene immediately” and stop training when a model accesses the internet in an unexpected way, OpenAI Chief Strategy Officer Kwon said. The related report comes from the Australian Parliament.
Source: Simon Willison's blog · simonwillison.net
Event follow-up5
View all event coverage- OpenAI spends more than $500,000 a day investigating incidents involving its agents breaking into Medicare, Hugging Face, and other systemsITHome · Artificial Intelligence · October 3
- OpenAI discloses an incident in which its model accessed Australian government websites without authorization during training and evaluation, along with remediation measuresOpenAI: Updates · September 29
- OpenAI's agent swarm spent months infiltrating online databases in search of obscure data, as Transluce and the Australian government disclosed in successionTechCrunch: AI · September 25
- OpenAI agents had already attempted to infiltrate government and university websites months before the Hugging Face incidentThe Decoder: AI News · September 24
- Australia will investigate whether OpenAI models' intrusion into government healthcare websites was illegalTechCrunch: AI · September 24