Plus: Anthropic’s latest model is facing backlash from some users. And Nvidia is partnering with medical note-taking company Abridge to develop a healthcare-focused AI model. Danny Lewis hosts.
Learn more about your ad choices. Visit megaphone.fm/adchoices
[00:00:00] [SPEAKER_02] Without trusted data, I don't think you have an AI advantage. You may actually be running an I-end liability at machine speed. That's Veeam's Shiva Pillay. Join him at the break to learn why trusted data is the fuel that lets AI scale successfully and safely.
[00:00:16] [SPEAKER_00] Here's your TNB Tech Minute for Thursday, June 11th. I'm Danny Lewis for The Wall Street Journal. We exclusively report that OpenAI is considering drastically lowering the prices it charges users as it seeks to win customers from Anthropic. People familiar with the matter say the move by the ChatGPT maker would be in anticipation of similar cuts expected by its rival. Business executives are beginning to balk at the high prices for AI usage as the two companies are trying to win over enterprise customers.
[00:00:43] [SPEAKER_00] But drastic price cuts could potentially erode the company's profit margins, which already lose billions of dollars due to the enormous costs of AI computing resources. News Corp., owner of The Wall Street Journal, has a content licensing partnership with OpenAI. Speaking of Anthropic, the Claude chatbot maker's latest AI model is facing backlash from some users over the broad restrictions on its capabilities. The company previously said its Mythos model was too dangerous to release widely because of fears around cybersecurity and bioweapon development.
[00:01:13] [SPEAKER_00] Some AI experts accused Anthropic of hiding some limitations to the update, called Claude Fable 5, including quietly degrading the quality of its responses about developing high-end AIs, while others found that Fable 5 refused to answer any questions about biology, mathematics, and chemistry. In a statement, Anthropic said it would make some safeguards visible and is working to reduce unnecessary obstructions.
[00:01:36] [SPEAKER_00] Plus, in another WSJ exclusive, we report that NVIDIA is partnering with Abridge, an AI note-taking app maker, to train a model focused on healthcare. The company said the AI is designed for clinical conversations and will be used exclusively within Abridge's platform to improve tasks like clinical decision support and documentation. As part of their work, Abridge will use de-identified clinical data to further train and customize NVIDIA's Nemotron suite of open models. The new model is expected to be ready for use later this year.
[00:02:06] [SPEAKER_00] And that's your T&B Tech Minute. We'll be back this afternoon with more.
[00:02:10] [SPEAKER_02] As companies deploy autonomous AI agents, the guardrails and contextual understanding provided by trusted data are critical. Here again is Shiva Pillay, GM and SVP at Veeam. Agentic AI doesn't wait for governance. Agents do not rest. The more bad data you're feeding to your models, the more potential harm. Trusted data is also essential to a company's ability to understand, govern, and precisely recover data at the same velocity and scale that AI can act on it.
[00:02:37] [SPEAKER_01] You need surgical precision for AI and data and data sets, particularly in time as they change. You isolate clean data, validate its integrity, and ensure nothing upstream is compromised. Recovery becomes prioritized by sensitivity tier, not by the size of a workload. So I'm not simply saying go and just recover a whole piece of data. Find me the sensitive data that I need to recover and why.
[00:02:58] [SPEAKER_02] Visit Veeam.com to learn more about how trusted data can power AI at scale. This content was created by custom content from WSJ, a unit of the Wall Street Journal Advertising Department.