Fearing leaks, Apple restricts ChatGPT use amongst staff

In line with inside sources and firm paperwork reviewed by The Wall Street Journal, Apple has restricted its staff’ use of ChatGPT and AI coding instruments equivalent to GitHub Copilot for worry of leaking confidential information to exterior sources. In the meantime, Apple can be reportedly creating comparable AI expertise.

ChatGPT is a conversational giant language mannequin (LLM) developed by Microsoft-backed OpenAI that’s able to duties starting from answering questions and writing essays to aiding with programming chores. At present, the ChatGPT AI mannequin solely runs on OpenAI or Microsoft servers and is accessible via the Web.

Apple’s determination to restrict using exterior AI instruments stems from a priority that confidential information might be uncovered as a result of nature of those AI fashions, which ship consumer information again to builders for cloud processing and help in future AI mannequin enhancements.

OpenAI beforehand confronted privateness points with ChatGPT when a bug permitted some customers to entry info from different customers’ chat historical past. In response to criticism from Italy, OpenAI applied an possibility for customers to turn off their chat history, which additionally prevents OpenAI from coaching future AI fashions utilizing their chat historical past information.

An ongoing pattern

Apple is well-known for its rigorous safety measures, and its determination to ban exterior AI instrument use displays a rising pattern amongst firms equivalent to Samsung and JPMorgan Chase. The Wall Avenue Journal reports that Amazon inspired its engineers to make the most of its proprietary AI instruments (like CodeWhisperer) for coding help as an alternative of exterior instruments.

To handle privateness considerations amongst companies, Microsoft recently announced plans to launch a privacy-focused model of ChatGPT that might goal banks, well being care suppliers, and different giant organizations with heightened considerations about information leaks and regulatory compliance.

This costlier enterprise model of ChatGPT is about to run on devoted servers, separate from these utilized by different firms or particular person customers, which is anticipated to assist forestall inadvertent information leaks and forestall delicate information from getting used to coach the AI language mannequin. Even so, it is uncertain a high-profile competitor like Apple would belief Microsoft with its confidential information beneath these circumstances.

Apple’s inside AI developments

The Wall Avenue Journal additionally says that Apple is working by itself LLMs beneath the management of John Giannandrea, a former Google worker who’s now Apple’s senior vice chairman of Machine Studying and AI Technique.

Lately, we covered an inside report from The Data that questioned the conservatism of Apple’s AI technique amid criticism of drawbacks in its now seemingly outdated Siri voice assistant. Siri, which launched in 2011, briefly inspired a miniature model of the AI hype we’re seeing now.

It is vital to notice that whereas Apple has restricted ChatGPT use for its staff, that ban doesn’t lengthen to Apple customers. Simply yesterday, OpenAI launched an official ChatGPT client for iPhone customers within the US, who can now simply entry the instrument on the go together with an OpenAI account.