Picture Credit score: iHeartMediaiHeartMedia is becoming a member of different firms like Spotify, Apple, and Verizon, warning their staff to not use ChatGPT.
An inside memo signed by iHeartMedia CEO Bob Pittman and CFO Wealthy Bressler outlined the expectations surrounding OpenAI’s new know-how. iHeartMedia instructed staff the usage of ChatGPT was to be severely restricted with out prior approval to be able to forestall leaks of proprietary data.
The e-mail outlines a sequence of pointers that prohibit ChatGPT use for iHeartMedia staff, together with use on firm units or importing firm paperwork to such platforms. iHeartMedia says the transfer was made to guard its mental property and different confidential data. That’s as a result of ChatGPT shops conversations it has with customers and makes use of these conversations to coach its AI algorithms. The massive language mannequin utilized in ChatGPT was skilled by scraping information from the web, Wikipedia articles, and different publicly out there on-line data.
“Though AI, together with ChatGPT and different ‘conversational’ AI’s will be enormously useful and actually transformative, we need to be sensible about how we implement these instruments to guard ourselves, our companions, our firms data, and our person information,” the memo to iHeartMedia staff reads. “For instance, for those who’re importing iHeart data to an AI platform (like ChatGPT), it would successfully practice that AI in order that anybody—even our opponents— can use it, together with all our aggressive, proprietary data.”
The memo outlines pointers for utilizing third-party AI instruments, together with chatting with iHeart’s authorized and IT groups earlier than implementing any use of the software. “All initiatives would require an evaluation of the enterprise affect and worth of the venture, a plan for monitoring and evaluating and a previous documented approval from Authorized and IT,” the memo continues.
iHeartMedia is only one of many firms looking for to guard its mental property from getting used to inadvertently practice LLMs like ChatGPT. Many of those firms have expressed issues that these conversations may very well be used to leak confidential data. An actual-world instance of this has already occurred when Samsung staff leaked commerce secrets and techniques by way of ChatGPT.