Apple Restricts Employee Use of ChatGPT and AI Tools, Report Claims
Meanwhile, the famously tight-lipped company is said to be working on its own large language model (LLM).
According to a report published by the Wall Street Journal (WSJ), Apple has implemented restrictions regarding how its employees use AI tools such as ChatGPT — not all that surprising given the Cupertino tech company is famously wary of leaks. The report from the WSJ suggests Apple’s employees won’t be disadvantaged by the lack of AI assistance forever, because Apple is busy creating “its own similar technology.” The report is based on an internal Apple document that appeared to back up these claims, as well as anonymous sources "familiar with the matter."
One of the issues with the use of AI based on large language models (LLMs) is that the tools usually use their interaction — inputs, questions, etc. — as training data. Samsung ran into this kind of problem back in April, when proprietary data was leaked from ChatGPT usage. The South Korean tech giant's fab engineers were using ChatGPT assistance for coding, note-taking, and data analysis about fab performance and yields, and three instances of data leaks were traced back to ChatGPT usage.
ChatGPT isn’t only prone to leaking other users’ data through training — like all software, it also has vulnerabilities and bugs. Earlier in the week we saw its vulnerability to prompt injection via YouTube transcripts. In March, ChatGPT was taken offline for a short period after it was discovered some of its users could see the titles of other user chat histories.
After the growth in reports concerning data security, OpenAI has introduced new privacy controls as well as an ‘Incognito mode’ for AI assistance on sensitive documents. Perhaps, these steps haven’t gone far enough for Apple. Another AI tool mentioned specifically in the WSJ report is GitHub’s Copilot (Microsoft), which is used by developers to save time with its smart code autocomplete functionality.
Not a lot is known about Apple’s progress on its own alleged LLM AI. The firm has snapped up a variety of AI related companies over recent years, and hired people such as John Giannandrea, an exec who used to work with Google's search and AI teams. Microsoft and Google are firing on all cylinders in the AI race, and are even pitching their AI-driven apps and widgets into Apple’s iOS store, so Apple can't be too patient with its own solution.
Apple’s developer conference, the WWDC23, kicks off on June 23, so perhaps it will have some AI news for us soon. There may be an AI-enhanced Siri announced, as well as highly anticipated Reality Pro AR/VR headset, new Macs, and new desktop, mobile, and wearable operating systems.
Stay On the Cutting Edge: Get the Tom's Hardware Newsletter
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Mark Tyson is a news editor at Tom's Hardware. He enjoys covering the full breadth of PC tech; from business and semiconductor design to products approaching the edge of reason.
-
hannibal Quite predictable. They don´t want to leak anything outside, so they make own AI...Reply
Nothing new in here. Move on, move on! -
Co BIY hannibal said:Quite predictable. They don´t want to leak anything outside, so they make own AI...
Nothing new in here. Move on, move on!
The important take away for the rest of us should be the realization that AI is already being used against the interests of the users.
If you aren't the paying customer then you are the product.