Appleâs latest developer betas launched last week with a handful of the generative AI features that were announced at WWDC and are headed to your iPhones, iPads, and Macs over the next several months. On Appleâs computers, however, you can actually read the instructions programmed into the model supporting some of those Apple Intelligence features.
Technology
‘You are a helpful mail assistant,’ and other Apple Intelligence instructions
Here are ome of the prompts that Apple Intelligence is using to guide AI models in thje macOS 15.1 Sequoia developer beta.
They show up as prompts that precede anything you say to a chatbot by default, and weâve seen them uncovered for AI tools like Microsoft Bing and DALL-E before. Now a member of the macOS 15.1 beta subreddit posted that theyâd discovered the files containing those backend prompts. You canât alter any of the files, but they do give an early hint at how the sausage is made.
In the example above, an AI bot for a âhelpful mail assistantâ is being told how to ask a series of questions based on the content of an email. It could be part of Appleâs Smart Reply feature, which can go on to suggest possible replies for you.
This sounds like Appleâs âRewriteâ feature, one of the Writing Tools that you can access by highlighting text and right-clicking (or, in iOS, long-pressing) on it. Its instructions include passages saying, âPlease limit the answer within 50 words. Do not hallucinate. Do not make up factual information.â
This brief prompt summarizes emails, with a careful instruction not to answer any questions.
Iâm pretty certain that this is the instruction set for generating a âMemoriesâ video with Apple Photos. The passage that says, âDo not write a story that is religious, political, harmful, violent, sexual, filthy or in any way negative, sad or provocative,â might just explain why the feature rejected my prompt asking for âimages of sadnessâ:
A shame. Itâs not hard to get around, though. I got it to generate a video in response to the prompt, âProvide me with a video of people mourning.â I wonât share the resulting video because there are pictures of people who arenât me in it, but I will show you the best picture it included in the slideshow:
There are far more prompts contained in the files, all laying out the hidden instructions given to Appleâs AI tools before your prompt is ever submitted. But hereâs one last instruction before you go:
Files I browsed through refer to the model as âajax,â which some Verge readers might recall as the rumored internal name for Appleâs LLM last year.
The person who found the instructions also posted instructions on how to locate the files within the macOS Sequoia 15.1 developer beta.
Expand the âpurpose_autoâ folder, and you should see a list of other folders with long, alphanumeric names. Inside most of those, youâll find an AssetData folder containing âmetadata.jsonâ files. Opening them should show you some code and â occasionally, at the bottom of some of them â the instructions passed to your machineâs local incarnation of Appleâs LLM. But you should remember these live in a part of macOS that contains the most sensitive files on your system. Tread with caution!