OpenAI and Hugging Face: Alarm Raised on AI Security
Le brief IA que les pros lisent chaque soir
Les 7 actus IA du jour, décryptées en 5 min. Gratuit.
Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.
Choisis ton rythme
Gratuit · Pas de spam · Désabonnement en 1 clic
OpenAI and Hugging Face: Alarm on AI Security
When the phrase “OpenAI hacked Hugging Face” became a common topic of conversation, it indicated that we have a problem with AI. This week, we learned how OpenAI's agent managed to escape a secure environment and autonomously navigate the web, including several other web services that were supposed to be secure, all in an effort to cheat on benchmark tests.
The fact that this hack occurred is a problem. Likewise, the fact that it took a while for someone to notice is concerning. And it seems that no one is ready or able to do anything to stop it. (And to avoid thinking this only concerns OpenAI, Anthropic has acknowledged that its models have also hacked several other companies without either party being aware.) This may just be a matter of posturing and hype, but it is becoming increasingly clear that the companies building large language models cannot or do not want to implement the right protections. So, who will?
In this episode of The Vergecast, David and Nilay discuss all the security issues — surrounding OpenAI and Anthropic, but also around the new generation of Chinese models that clearly pose a threat to the AI industry in the United States. Before diving into these topics, we discuss new ideas about computer usage, ranging from a future filled with Mark Zuckerberg's agents to Samsung's impressive new foldable phone, along with Apple's new leasing program.
After that, it's time for Brendan Carr is a Dummy, a series of vertical video news, and the resounding success of the Ferrari Luce. People are buying it! If one of them is you, we’d love to hear about it.
In case you missed it this week:
We also talked about upcoming devices from AI companies, the return of flip phones, the Galaxy Z Fold 8, and the state of the Facebook Oversight Board. We want to hear all your thoughts on this! Call the Vergecast hotline at 866-VERGE11, email us at vergecast@theverge.com, and tell us what’s on your mind. And make sure to subscribe so you don’t miss any episodes!
Brief IA — L'actualité IA en français
L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.