When AI Agents Cross the Boundary: Securing the AI Factory
AI Proving Ground Podcast: Exploring Artificial Intelligence & Enterprise AI with World Wide TechnologyShow Notes
In July, OpenAI disclosed that models undergoing a cybersecurity evaluation escaped their testing environment, exploited a zero-day vulnerability and ultimately reached Hugging Face’s production infrastructure. The incident raised a difficult question for every organization deploying autonomous AI: What happens when an agent crosses a boundary — not because it was instructed to attack, but because doing so helps it complete the objective it was given? Recorded before the incident became publi...
More from AI Proving Ground Podcast: Exploring Artificial Intelligence & Enterprise AI with World Wide Technology
For years, enterprises have been stuck with aging software that's too expensive, too complicated or too risky to replace. AI is changing that equation. But faster code doesn't necessarily mean better software. What happens when AI can rebuild a critical application in a fraction of the time, but the people who understood why it worked that way are long gone? On this episode of the AI Proving Ground Podcast, WWT's Nate McKie joins host Brian Feldt to explore how generative AI is rewriting the ...
Is your enterprise ready for AI it can't fully control? WWT CTO Mike Taylor breaks down the week's biggest AI news: Bank of America's patching push, Satya Nadella's call for an AI emergency brake, Anthropic's warning about AI that copies itself and the CoreWeave debt jitters. In this episode of The Weekly Signal, host Brian Feldt and Mike Taylor dig into the trust gap behind this week's enterprise AI headlines. The topics include cybersecurity, AI agents, AI governance and GPU infrastructure...
For decades, network engineers have been the ones keeping everything running. Now AI is learning to do the job itself. And in some cases, it's already doing it better. As AI agents place unprecedented demands on enterprise networks, the infrastructure supporting them is undergoing its own transformation. Networks that can detect problems, optimize performance and even repair themselves are moving from ambitious concept to operational reality. On this episode of the AI Proving Ground Podcast, ...
AI-powered cyberattacks are getting faster, billions are pouring into compute infrastructure and the promised productivity gains are still difficult to measure. Is enterprise AI moving faster than businesses can adapt? WWT CTO Mike Taylor joins Brian Feldt to unpack the week's biggest AI developments, from Anthropic's expanding cybersecurity efforts to the growing debate over compute capacity and AI's economic payoff. They explore why stronger security requires more than better tools, how org...
AI is getting cheaper, faster and easier to put to work. So why are so many companies still struggling to show a return? Purdue University’s Richard Ryffel and Mohammad Saifur Rahman join the AI Proving Ground Podcast to argue that the problem may be how businesses are thinking about AI in the first place. Treating it as another efficiency tool can only take you so far. The bigger opportunity comes when companies use cheaper intelligence to rethink how work gets done, where humans create valu...