OpenAI CEO Sam Altman pushed for enterprise-level cyber defence solutions against AI-enabled attacks in light of the Hugging Face incident, where his own company’s AI models hacked into another company’s infrastructure.
During a free-wheeling conversation with Salesforce CEO Marc Benioff during the first day of the Dreamforce enterprise technology event, Mr.
Altman shared more background details about his company’s rogue AI model attack against Hugging Face and the need to pace AI progress.
Trump says there is ‘SICK conspiracy’ against AI and data centres, dismisses new guardrails “These models have gotten so good, so fast, that we need to treat the alignment and monitoring security with a new level of rigour, and we have to be willing to pace our development of capabilities, such that [...] alignment, safety, and monitoring are always ahead of capabilities.
And the Hugging Face incident triggered a real reset, not just in our own company, but I think the whole industry,” explained Mr.
Altman.
At the same time, he pushed back against the idea of keeping certain advanced AI models inaccessible for enterprise use.
“But more generally, we want to be a great dependable partner to enterprises, and we don’t want to be, like, ‘We’ve got this great model, we’re going to keep it locked up, and not let you use it,’” he explained.
Mr.
Altman stressed on the need for enterprises to have a defence against a new wave of AI-enabled cyber-attacks, but said he did not believe smaller companies were at a disadvantage when it came to cybersecurity.
“First of all, I think there are probably more advantages to being a smaller number-of-employee company than ever before.
The speed with which you can move, the way that you can adopt new technology, the way that you can, like, be ahead of the curve on this stuff,” observed Mr.
That being said, he acknowledged there was a personal motive as well.
Experts have pointed out that even as AI companies such as OpenAI and Anthropic disclose and discover instances of their own AI agents going rogue, the same companies are the ones offering cyber-defence solutions as well.
How OpenAI’s rogue agents took over message boards | Explained “And then, you know, selfishly, we would love to sell you Daybreak services to help defend against attacks,” said Mr.
Altman, referring to OpenAI’s cyber-defence stack.
“Our model hopefully will not be attacking,” he added, when Mr.
Benioff half-jokingly asked whether this defence was against OpenAI models or other models.
However, the OpenAI CEO did end on a note of optimism, predicting that AI users’ personal and professional lives would drastically improve in the next five years.
