Abstract:
An OpenAI spokesperson revealed that OpenAI President Greg Brockman will soon make an announcement to expand the authority to open the company's AI tools to critical infrastructure and public sector agencies. The accidental intrusion of Hugging Face by OpenAI, coupled with multiple cyberattacks on U.S. water systems suspected of using AI, has aroused great concern from the outside world about how vulnerable various public facilities are to future AI-driven hacker attacks.
The spokesperson said that Brockman will officially announce the above information at the Cyber Security Summit held at OpenAI headquarters on Thursday.
About 300 corporate security leaders and chief information security officers from Fortune 100 companies are expected to attend this meeting.
The core topic of this summit: How enterprises, governments, and network security vendors enable defenders (critical infrastructure agencies, public service departments) to use AI tools.
This meeting and the upcoming announcement are OpenAI’s response to model security risks. Hugging Face has been hacked before, and OpenAI’s upcoming Astra model already has “critical level” cyber attack capabilities.
Earlier this week, a report mentioned that OpenAI was developing new training technology, which also caused concerns in the industry: whether security practitioners can still see the reasoning logic of the model clearly.
OpenAI issued an open letter last week, warning that organizations have only months to respond to the threat of AI-enabled cyberattacks.
Brockman himself issued a similar warning in an article last month.
Raising awareness of critical infrastructure protection is only the first step, and it cannot solve the practical problems of insufficient funds and talent shortages in security teams commonly faced by these institutions.
Comments