# ChatGPT Can Access iMessage and Meta Glasses Detectors

An episode of [AI Chat: AI News on Claude, ChatGPT, Gemini, Grok, Perplexity](https://invisiblenetwork.ai/podcast/1669990482), 22 Aug 2026, 15 min.

Audio: https://content.rss.com/episodes/365073/3088204/ai-chat-podcast/2026_08_22_02_33_15_df52d12a-5a84-4fbe-9b7e-f2265d07f07c.mp3

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! In this episode, we discuss ChatGPT's new capability to access Apple’s iMessage, exploring its potential uses and the privacy concerns that arise. We also cover the rise of AI-related communities and companies, including the growth of Micro One and innovative developments like Every's AI clone of their editor-in-chief. Chapters 00:00 ChatGPT and iMessage 02:01 Privacy Concerns 05:00 Detection Apps and Meta Glasses 07:59 AI Research Accelerating Concerns 09:59 Micro One's Growth 12:59 Every's AI Clone Innovation Show Links • Get the top 80+ AI Models for $8.99 at AI Box: https://aibox.ai • How I Grow and Scale My Business with AI: https://www.skool.com/aihustle • Get the AI Chat Daily Newsletter: https://www.aichatdaily.com/newsletter

## Transcript

[0:00] ChatGPT is getting access to Apple's iMessage
[0:03] so it can draft, send texts, search through all
[0:06] your messages. There's a lot of people concerned
[0:08] about the security implications, but also a lot
[0:11] of people wondering what you can actually do
[0:12] with this. We're going to get into that on the
[0:13] podcast. Also, a bunch of detector apps are getting
[0:17] very popular as meta AI glasses have this huge
[0:21] privacy backlash. So we'll get into that. And
[0:23] right now there are 1100 labs and AI employees
[0:27] that are asking you. governments to help pace
[0:30] automated AI research. They're worried that it's
[0:33] going to get out of control and be hard to actually
[0:35] control what's going on when AI is teaching itself.
[0:38] Micro One has just hit $500 million in run rate
[0:42] as AI training demand is getting incredibly popular.
[0:45] This is a business, an area that I personally
[0:47] am working on. So I have a lot of... insights
[0:50] into this. And also every builds AI clone of
[0:54] their editor in chief from 30 ,000 edits. So
[0:57] every is a publication and they built basically
[0:59] a clone. They had their editor go and make edits
[1:02] to all of the journalism, all the work that was
[1:04] getting published to their platforms. And then
[1:06] they said, hey, look at the document beforehand,
[1:09] look at the edits that they made and let's train,
[1:12] basically let's tell the AI how to edit things.
[1:15] And after that, they were able to scale up from
[1:17] 30 to 60 people. very quickly and actually have
[1:20] the AI doing a lot of the editing for them. If
[1:23] you want to get access to different tools and
[1:26] strategies on actually growing and scaling your
[1:28] business with AI, I mentioned this a couple of
[1:30] days ago on the podcast. I wasn't planning on
[1:32] talking about this for another week, but my AI
[1:35] Hustle School community is growing super fast.
[1:38] And I mentioned in the past, once we hit a thousand
[1:39] members, we're going to pay wallet and make it
[1:41] paid only. And it is currently at... 928 members.
[1:45] I wasn't going to mention it for a week, but
[1:47] it grew like a couple hundred members in the
[1:49] last couple of days. So we're going to get very
[1:51] close to a thousand very fast. If you're interested
[1:54] in joining the AI Hustle School community, it's
[1:56] completely free to join right now. You get access
[1:58] to over 900 people that are building. I'm going
[2:00] to cover how you can. grow and scale your business
[2:03] using AI tools, but also how to build out software
[2:05] and apps, all the stuff that I have done everything
[2:07] from marketing to actually developing the tool.
[2:10] So if you're interested in growing and scaling
[2:12] your business with AI tools, go check it out.
[2:14] There's a link in the description to the AI hustle
[2:15] school community, it is free to join right now.
[2:17] But as soon as we hit 1000, we're going to pay
[2:19] wallet, it's going to be pretty expensive, not
[2:21] expensive, it's going to be an incredible value,
[2:22] but it'll be probably somewhere between 20 or
[2:25] $50 a month, depending on what kind of promotions
[2:27] are running. So if you want to get it free, I
[2:28] would go get it now. Alright, let's talk about
[2:30] what's going on with chat GPT. So like I mentioned,
[2:33] it just got access to iMessage. And this is actually
[2:36] via a plugin. So I originally launched a plugin
[2:38] for chat GPT. And then they also have let it,
[2:42] they've also added it to ChatGPT work and Codex.
[2:45] So basically all of the OpenAI, all of the OpenAI
[2:48] tools can access your iMessage if you allow it.
[2:51] Of course, you have to go and allow it. They
[2:52] can sort, it can read, it can draft, and it can
[2:55] send iMessages. Now, there's a bunch of places
[2:58] that my mind goes to immediately. One of them
[2:59] is that there's a ton of tools, one in particular
[3:02] called SendBlue, where like for... salespeople
[3:05] and cold outreach, they use these different specific
[3:07] tools that will allow it to basically look like
[3:10] you're texting from an iPhone. And they do this
[3:13] so that when someone gets a text on an iPhone,
[3:15] if it looks like it's coming from an iMessage,
[3:17] they have a higher perceived trust versus if
[3:19] it's a green text message from any other sort
[3:22] of SMS sender. So to me, immediately my mind
[3:25] went there because that's an interesting industry
[3:27] and an interesting business. And now you essentially
[3:29] could automate that, let ChatGPT have access
[3:32] to your iMessage and you could probably have
[3:34] access to your businesses, quote unquote, iMessages
[3:36] that you would have on your computer. And you
[3:39] could get, you know, verified blue bubble text
[3:41] messages for free. So that's like one interesting
[3:44] benefit. But I think a lot of people are concerned
[3:46] about the privacy element. I think even they
[3:50] were kind of warning, opening eye warned against
[3:52] enabling persistent approval, which is what I
[3:55] usually give ChatGPT and Claude when it's working
[3:57] on something. I'm like, look, you get unlimited
[3:59] approval, just do whatever you need. And, you
[4:01] know, it still asks me like if it has to copy
[4:03] and paste a password, it's going to make me do
[4:05] that. But for a lot of the tasks, I don't want
[4:07] it to ask me every five seconds, like the direction
[4:09] of the task it's taking. But apparently they're
[4:12] warning in this case, it's not great because
[4:13] then it's going to remove your final chance to
[4:16] review anything that you send. Now, if you're
[4:18] using this to like organize your life and to
[4:20] reply to your grandma, you probably want to,
[4:22] you know, review what's being sent here. But
[4:25] if this is something that you're using for maybe
[4:27] work or maybe kind of like sales and you're doing
[4:29] high volumes, 100%, this is something that I
[4:32] would use. I told Bloomberg that the plugin runs
[4:34] locally on your device, so it doesn't create
[4:37] an index of all of someone's messages. All of
[4:39] that is happening on your device. A lot of people
[4:41] are worried about the privacy concerns of this.
[4:43] It's, like I mentioned, going to work with Codex
[4:45] and ChatGPT. work. And they had this kind of
[4:48] launch demo, they basically showed ChatGPT drafting
[4:50] some follow up messages based on the priorities
[4:53] conversations and searching deep message history.
[4:56] So honestly, that's kind of useful if it's going
[4:58] to go down a deep rabbit hole and get some information
[5:01] from a chat you had a long time ago, ChatGPT
[5:04] can go and do the research for you, which is
[5:06] pretty useful. The thing that I think is interesting
[5:08] here is if Apple had really integrated proper
[5:11] Siri integrations and you know the quote -unquote
[5:14] apple intelligence had really worked none of
[5:17] this would actually be necessary you'd just be
[5:18] using apple's onboard you know whatever built
[5:21] -in software and tools but because they didn't
[5:24] do that now there's kind of these things popping
[5:26] up from popping up from apple there's a whole
[5:30] bunch of detector apps that are getting very
[5:32] popular as people are having a little bit of
[5:34] backlash against these new smart sunglasses or
[5:38] smart glasses the meta ray bands and other meta
[5:40] glasses that are coming out and there's an app
[5:42] called nearby glasses it's three dollars and
[5:45] you can get on ios and android and it is at about
[5:47] 100 ,000 downloads on the Play Store. This just
[5:50] came out in February. There's another one called
[5:53] Zuck Off that launched. It's free on iPhone.
[5:56] It just launched, and it is a $10 a year perpetual
[5:59] price for it. But essentially what these things
[6:02] do is they're actually scanning the Bluetooth
[6:04] of all the surrounding devices, and anytime a...
[6:08] device that is, you know, a meta Ray Bans, or
[6:11] even if it's something like the Snapchat spectacles,
[6:13] if there's any of these smart glasses that pop
[6:15] up on the radar, if they hit them with the Bluetooth,
[6:18] they're going to notify you and let you know
[6:20] that look, there's probably some meta Ray Bans
[6:22] around you that could be recording you. So I
[6:24] think a lot of people enjoy like our are happy
[6:27] with that meta installed and then they also unreleased
[6:30] some facial recognition systems on about 50 million
[6:33] phones in june they actually deleted it because
[6:35] wired and eff kind of exposed it and talked about
[6:39] it so they went and removed that but at defcon
[6:41] 2026 they actually banned smart glasses completely
[6:44] there's a bunch of schools courts restaurants
[6:46] entertainment venues that have also banned smart
[6:50] glasses there's another app called anti -zuck
[6:52] smart glasses scanner that got to very close
[6:56] to the top of the iphone charts in july it was
[6:58] three dollars so they're kind of all in that
[7:00] same all in that same ballpark so ives jean -marc
[7:04] who described himself on github as a no pro developer
[7:07] warned that detecting bluetooth le devices might
[7:10] sometimes just not work as expected and false
[7:12] positives are likely so To be fair, these aren't
[7:14] always accurate. And I think that might be kind
[7:17] of worrisome for some people if you're laying
[7:18] in bed one night and all of a sudden your phone
[7:20] beeps that there's smart glasses and you look
[7:22] around and there's no one near you. That would
[7:24] be kind of an awkward false positive. Those do
[7:26] happen. I don't know how likely that is, but
[7:29] it is important to be aware of. There are 1 ,100
[7:33] AI lab employees that are asking the US government
[7:39] to help pace automated AI research. There's a
[7:42] bunch of people from... OpenAI, Anthropic, Google,
[7:44] and Meta, essentially what they're concerned
[7:46] about in particular is that automated AI research
[7:49] could accelerate beyond human control without
[7:52] coordinated governance. Now, it seems kind of
[7:55] like a... perhaps a regulatory capture approach
[7:58] because this is OpenAI, Anthropic and Google
[8:00] and Meta. These are the biggest companies that
[8:02] are going to have a lot of budget to spend on
[8:05] this kind of stuff. They have the most money
[8:07] to spend, you know, like hundreds of billions
[8:09] of dollars or in the case of Anthropic, a billion
[8:13] dollars a month just on, you know, data centers
[8:15] for data processing and for running their models.
[8:19] So some of these companies have an insane amount
[8:21] of money. And so when they're going to the government
[8:24] and saying, hey, look, we're worried that like
[8:25] Like our training is going to accelerate beyond
[8:27] human control. You need to put a bunch of like
[8:29] governance in place. I'm a little bit pessimistic
[8:32] because it feels like they're trying to pull
[8:33] up the ladder behind them. These aren't small
[8:35] companies asking for this. These are the big
[8:37] companies that have big budgets that, you know,
[8:39] theoretically, they're the ones that could mistakenly
[8:42] have their training model perhaps run and accelerate
[8:46] beyond what they're capable of doing. There's
[8:48] also, I think, Microsoft Mistral and Thinking
[8:50] Machines that signed the pacing the frontier
[8:53] statement. All of the people were basically asking
[8:56] the government to back an internal effort to
[8:58] build tools that can deliberately slow automated
[9:00] AI research. OpenAI had their CRO, Mark Chen,
[9:04] and chief scientist, Jacob Paco, and also their
[9:08] co -founder, John Schulman. and some others sign
[9:11] it. Anthropic had their co founder Jack Clark
[9:13] sign it. So there's a lot of people that are
[9:15] working on this right now. The statement follows
[9:18] last week's incident where you had opening I
[9:20] escaped in the sandbox hacking hugging faces
[9:23] or maybe the week before. So obviously, this
[9:25] is something that people are concerned about.
[9:26] They're like, look, if if the AI models out there
[9:28] hacking things, we don't want it to, you know,
[9:30] progressively get better and better. In my opinion,
[9:32] you can always unplug the machine. I know people
[9:36] probably hate that statement, but at the end
[9:38] of the day, I think that I don't always love
[9:40] a lot of this. regulation that comes out, perhaps
[9:43] some of it's necessary, but it feels like there's
[9:45] a bunch of questions that are not being answered
[9:48] in this. It doesn't specify how pacing would
[9:51] be verified across jurisdictions, how China based
[9:53] labs would be brought into any coordination framework.
[9:56] And the spoiler alert is they wouldn't be. So
[9:59] all of these US companies are going to decide
[10:01] that, hey, there's this kind of like coordinated
[10:03] framework, we're not going to let the AI model
[10:05] train itself to make itself better and faster.
[10:07] And do you think the Chinese Open source models
[10:09] are going to care? Absolutely not. They're going
[10:10] to just run with that. So I think at the end
[10:12] of the day, this probably isn't going to pass
[10:14] and it probably isn't super viable because of
[10:16] that. But that's just my opinion. Micro One has
[10:18] just hit $500 million in run rate as AI training
[10:22] data demand is surging. This is a company that's
[10:25] only four years old. It's a data labeling company
[10:27] and they grew fivefold in eight months. So they're
[10:30] right behind Merkur, who is at $2 billion and
[10:33] Handshake is at $1 billion. Both of all three
[10:36] of those companies are in the same market. market.
[10:37] Micro one's gross annual run rate grew from about
[10:40] $100 million to $500 million over the last eight
[10:44] months. That is a five X growth rate. That is
[10:46] absolutely insane. The net run rate lands between
[10:49] about 150 million and 200 million after 60 %
[10:53] to 70 % contraction. contractor retention. Merkur
[10:56] hit $2 billion in gross annualized revenue this
[11:00] summer and Handshake reached a billion dollars
[11:02] earlier this year. So what's interesting to me
[11:04] is what these companies are actually selling.
[11:07] Off the shelf data sets sold to multiple customers
[11:09] are about 80 to 90 % gross margin. And MicroOne
[11:13] raised their Series A at a $500 million valuation
[11:16] last September. And they might have also recently
[11:19] raised again at a higher market, although it
[11:21] wasn't exactly. disclose. This is actually an
[11:24] area that I'm working in. I'm specifically working
[11:26] with podcasters and people that have video interviews.
[11:30] I have a bunch of companies that I work with
[11:32] where we license data to them and they'll purchase
[11:36] data sets of essentially people talking in high
[11:38] quality audio and video to help train general
[11:41] purpose. It's not making video or voice clones
[11:43] of people, but it's helping train AI models on
[11:46] how to talk, how conversations happen. They sell
[11:48] this to a number of different AI companies. And
[11:50] it's actually... a very lucrative, especially
[11:52] if someone has a big data set of a bunch of recordings
[11:55] or interviews that they've done over the years.
[11:56] It's a really great way to make money from something
[12:00] that you already have. I think a lot of people
[12:01] are realizing this and it is a massive growing
[12:03] industry. Now, a lot of these companies here
[12:05] aren't just selling the data, but they're also
[12:07] labeling it. They're also processing the data.
[12:09] There's a ton of work that has to go into actually
[12:10] making this data usable for AI companies when
[12:13] they're training their models. And so there's
[12:15] a lot that goes into it. But at the end of the
[12:18] day, It's just an absolutely massive market that
[12:21] is growing very, very quickly. A company called
[12:23] Every has built an AI clone of their editor -in
[12:26] -chief and they did that using... 30 they trained
[12:29] it on 30 ,000 edits. Their CEO is Dan shipper.
[12:32] And he says that they have a 30 person team.
[12:35] And they were able to double their headcount
[12:37] while they automated all of their copy desk,
[12:40] they built an agent, Kate Lee was their editor
[12:42] in chief, and they built an agent that cloned
[12:45] all of Kate Lee's editing, which is very wild.
[12:49] The CEO said we have an editor in chief, Kate
[12:51] Lee, who's fantastic, who I've been trying to
[12:53] automate out of a job for years in an extremely
[12:56] benevolent way. I mean, I think they're benevolent
[12:58] because she has a lot of work to do but every
[13:02] single collected one of her 300 or of her 30
[13:06] 000 historical edits and from being an editor
[13:09] -in -chief they were used to train an internal
[13:11] copy editing agent that back tests against her
[13:15] past work so it will go and edit something and
[13:18] then it will go check you know based off her
[13:19] past work would this Would this be approved?
[13:22] Would this work? The company doubled, like I
[13:24] mentioned, from about 15 people to about 30 people
[13:26] over the last year. They're running six software
[13:29] products and they also have a daily newsletter
[13:31] and they have kind of a bundle of, you know,
[13:33] people subscribe to a bunch of their different
[13:35] journalist endeavors. They're able to launch
[13:38] in 2020 and they haven't raised any or haven't
[13:41] raised a lot of venture funding from that. But
[13:43] I just think this is an interesting place. You
[13:44] know, I obviously have. aichatdaily .com, which
[13:47] is my own news site where you can get deep dives
[13:49] on all of this content that I'm talking about
[13:51] here. I have articles breaking down way more
[13:54] information on all of these stories. So if you're
[13:55] ever interested in getting that into your inbox,
[13:57] there's actually a newsletter on aichatdaily
[13:59] .com where you can go subscribe and get all of
[14:02] these stories and deep dives into your inbox
[14:04] as a newsletter. But I use a very similar process
[14:07] where I have AI agents that are editors, writers.
[14:11] reviewers, we go and source content to write
[14:14] about. And I think so much of the news process,
[14:18] there's so many ways to help automate real people
[14:20] that are doing real work in this area to help
[14:23] speed up some of the slower parts. And I mean,
[14:27] in the case of this one poor editor who's trying
[14:29] to edit for 30 people, I'm sure she's thrilled
[14:31] to get a little bit of her time back. Thanks
[14:34] so much, everyone, for tuning into the podcast
[14:35] today. Make sure if you're interested in getting
[14:37] access to the AI Hustle School community. Once
[14:40] it reaches a thousand users, we're going to make
[14:41] it paid. It's free to join right now. So if you're
[14:44] interested, go check it out. It's linked in the
[14:46] description, how to grow and scale your business
[14:48] using AI tools. And I'll catch you guys all in
[14:51] the next episode.

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.