# Anthropic Faces FTC Scrutiny as OpenAI Claims a Major Math Breakthrough

An episode of [AI Chat: AI News on Claude, ChatGPT, Gemini, Grok, Perplexity](https://invisiblenetwork.ai/podcast/1669990482), 27 Oct 2026, 16 min.

Audio: https://content.rss.com/episodes/365073/3272176/ai-chat-podcast/2026_10_27_23_35_54_bd00a0b3-0ee9-46eb-8667-85f7845c198d.mp3

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! • Anthropic faces new scrutiny as the FTC seeks executive testimony in its consumer-harm investigation. Jaeden breaks down the news, the reported $63 billion financing plan for Broadcom chips destined for Anthropic, and OpenAI's claim that an internal model refuted the Exponential Time Hypothesis. We also cover Verizon's $2 billion-plus AWS agreement, GPT-6.1 Terra, and Oracle's new Fusion Claw agentic applications. The math result is a claim under review—not a solution to P versus NP—and the FTC investigation is not a finding of wrongdoing. Join the AI Hustle Skool community for practical AI business workflows: https://www.skool.com/aihustle Sources discussed: FTC testimony report: https://www.axios.com/2026/10/26/ftc-ai-safety-probe-anthropic-openai OpenAI mathematical research: https://github.com/openai/math Verizon earnings and AWS agreement: https://www.verizon.com/about/news/verizon-delivers-another-record-breaking-quarter

## Transcript

[0:02] Welcome to the AI Chat Podcast. I'm your host,
[0:04] Jaden Schaefer. Every day I cover cutting -edge
[0:07] AI news and talk with the leaders behind it,
[0:09] breaking down what it means for your life and
[0:11] business. Welcome to the podcast. The FTC is
[0:18] demanding that anthropic executives testify in
[0:22] their AI consumer harm probe. We've seen this
[0:24] in a lot of different industries with social
[0:27] media, and now it is moving into AI. We have
[0:31] banks and Blackstone that are lining up. 63 billion
[0:35] dollars right now to fund broadcom ai chips and
[0:38] these are all going to anthropic we also have
[0:41] opening eye that says their internal model disproved
[0:44] the exponential time hypothesis this is my science
[0:47] corner that i'm excited to cover today on the
[0:49] show We have Verizon that is signing a $2 billion
[0:52] AI deal with AWS, and they're raising their 2026
[0:55] Outlook, which a lot of people are excited about.
[0:58] OpenAI is launching GPT 6 .1 Terra. It's going
[1:01] to be half the price of GPT 6 .1 Sol. And there's
[1:04] a bunch of interesting reasons why that's going
[1:06] to make a difference. Oracle has launched 25
[1:09] Fusion Claw agentic apps across ERP. HCM, supply
[1:15] chain, and CX. So we're going to get into what
[1:17] Oracle is launching there and a couple other
[1:19] interesting things. Before we get into all of
[1:21] that, I wanted to mention, if you have ever wanted
[1:23] to start a business with AI, create a side hustle,
[1:26] or grow and scale your current business or your
[1:28] career, I'd love for you to go check out the
[1:31] AI Hustle School Community. We have 1300 members
[1:33] and I make a new video here every single week
[1:36] breaking down how I'm scaling and growing my
[1:38] businesses with AI. We have a classroom section
[1:41] with over 100 different videos. And my goal basically
[1:43] is for people to join the school community and
[1:45] make an extra $10 ,000 a month from a side hustle
[1:48] or business. If you have a much bigger business,
[1:50] you can grow that a lot. But if it's a new venture,
[1:53] something new for you, that's my goal. We have
[1:55] a couple different courses. Essentially, we show
[1:57] you how to build and grow SaaS apps. We show
[1:59] you how to become an AI consultant. We talk about
[2:02] how I make money doing AI generated music, podcasting.
[2:06] And if you already have a product, we go over
[2:08] all of the strategies for organic growth, social
[2:11] media growth, paid ads, offers, emails, newsletters,
[2:14] all that in. And I think we have an archive of
[2:17] over 100 different videos breaking down different
[2:19] tools, strategies, techniques. I would love for
[2:21] you to join. I have. personally created apps
[2:24] that have over 150 ,000 users. And I recently
[2:27] vibe coded an app that just passed a thousand
[2:28] users as a brand new app. I talk about all the
[2:31] strategies in there and a lot of other cool stuff.
[2:33] So it's only $19 a month to join. We made this
[2:35] very cheap. If you're interested, go check it
[2:37] out. There is a link in the description. All
[2:40] right, let's talk about what's going on with
[2:41] the FTC right now. So they have demanded the
[2:45] anthropic executives are going to come and testify
[2:47] for their probe. The testimony request follows
[2:50] a 50 -page document demand sent to Anthropic
[2:53] and OpenAI. And of course, everyone knows this
[2:55] is happening right as Anthropic is pitching their
[2:57] investors on their IPO. They've already filed
[3:00] for their IPO. OpenAI said that they're not going
[3:02] to file till next year. So we have these two
[3:04] different timelines going on right now. Anthropic
[3:07] and OpenAI have both each received this 50 -page
[3:10] civil investigation demand. They're talking about
[3:13] a bunch of different things here. One of them
[3:14] is security incidents. One of them is safeguards.
[3:16] One of them is marketing. And one of them is
[3:18] privacy. an FTC spokesperson said that individuals
[3:21] at other AI companies have also received testimony
[3:24] demands. They didn't say what companies, but
[3:26] we know for sure we got Anthropic and OpenAI
[3:28] that are requested to come and testify. And we
[3:30] probably have a bunch of others. And I think
[3:32] there's a whole bunch of different things that
[3:33] we'll see here. There's kind of the usual lawsuits
[3:36] that we've seen around like teen safety and different
[3:38] things like that. But I think at the same time,
[3:39] we're also going to see, you know, when they
[3:41] talk about marketing, it makes me think of what
[3:43] some people have accused Anthropic of doing with
[3:45] marketing stunts where they'll go and find security
[3:48] vulnerabilities. I mean, it's an interest. It's
[3:50] so tricky because some of these security vulnerabilities
[3:52] are real that Anthropic is discovering and they're
[3:55] telling people about. But at the same time, it's
[3:57] just a lot of people have criticized them saying
[4:00] it comes off as wild that they discover a security
[4:02] vulnerability that requires millions of dollars
[4:06] worth of tokens to go fix. And then they tell
[4:08] all the companies and all the companies have
[4:09] to use their models to go in. fix all the security
[4:11] vulnerabilities. So they make money discovering
[4:14] them, make money from the patching. But at the
[4:16] end of the day, people have bounty programs for
[4:18] these discoveries. And these are some really
[4:20] big security issues that have probably existed
[4:22] for a very long time. So, you know, obviously
[4:25] there's a lot of good, but it's going to be a
[4:27] pretty wild investigation to see how much of
[4:29] this is, you know, legit, everything, everything
[4:32] that's going on. Anyways, the demands went out
[4:34] the same day the Anthropic filed their IPO. So
[4:37] it's definitely at a tense place for the company.
[4:40] Banks and Blackstone right now are lining up
[4:42] $63 billion to fund Broadcom. These are going
[4:45] to be chips that are going to Anthropic. It's
[4:48] a new special purpose vehicle, an SPV, that is
[4:51] going to buy Broadcom XPU accelerators and networking
[4:55] gear, and it's going to basically lease them
[4:57] to Anthropic. The deliveries are starting in
[4:59] 2027. It's interesting because we have to get
[5:02] creative. Anthropic's already raised so much
[5:04] money. They're about to IPO. Broadcom is obviously
[5:07] a... a company that is very well known. But even
[5:10] with a company like Broadcom and a company like
[5:12] Anthropic, making it just a clear deal for Anthropic
[5:14] to go acquire chips from Broadcom is sometimes
[5:16] tricky because of the way that these deals are
[5:18] structured. And so we're able to do all sorts
[5:21] of... creative financial things like spinning
[5:23] up new SPVs to go and actually acquire the chips
[5:26] and lease them. So it's not like Anthropica is
[5:28] buying them directly from Broadcom for basically
[5:31] financial reasons. Bank of America, Blackstone
[5:33] City, Morgan Stanley, all of them are creating
[5:36] this new SPV. And Broadcom is going to give residual
[5:40] value support to it, which is equal to the first
[5:44] lien portion of the debt. The way debt works,
[5:46] the way lending works, the way the terms work,
[5:49] this is kind of the best way for Anthropic to
[5:51] get its hands on these chips. Deliveries are
[5:52] expected to start next year under the Compute
[5:55] Capacity Agreement that Broadcom announced earlier.
[5:58] Apollo and Blackstone previously agreed to finance
[6:00] a $35 billion extension to Anthropic's computing
[6:03] capacity. And so now we kind of have these follow
[6:06] -on companies that are... that are adding more
[6:08] to this. All right, we have a big one. This is
[6:11] a science corner for me, but basically OpenAI
[6:13] says that their internal model has disproved
[6:16] the exponential time hypothesis. So I found a
[6:19] really good tweet over on X that explains the
[6:21] exponential time hypothesis, and I want to read
[6:24] it to you. It's not too long, but I just feel
[6:27] like this is a really interesting problem that
[6:29] computer science has been grappling with for
[6:31] a super long time. and through our OpenAI is
[6:34] able to solve it, there's a bunch of really interesting
[6:36] implications. So we had Greg Brockman that was
[6:39] actually tweeting about having this solved among
[6:41] a bunch of other ones, right? So we had a whole
[6:43] bunch of, you know, math problems that were solved
[6:46] a while back by OpenAI. And apparently there's
[6:48] a whole bunch of new ones that they've now just
[6:49] done. So anyways, there is a really interesting
[6:51] tweet that went out all about this by Ramesh
[6:55] Karnawat. And they said, many of you don't understand
[6:58] how serious this is, so let me break it down.
[7:00] In computer science, there's a famous kind of
[7:02] puzzle called the three sat. You're given a long
[7:05] list of yes or no switches and a set of rules
[7:08] about how they must be arranged. And you have
[7:10] to find a combination that satisfies every rule.
[7:13] The obvious way to solve it is to try every combination.
[7:16] But each new switch that you try doubles the
[7:19] number of possibilities. So a puzzle with a few
[7:22] hundred switches would take longer than the age
[7:24] of the universe to brute force, right? So it's
[7:26] like, you know, here's a hundred switches, get
[7:29] them all in the right order and, you know, got
[7:30] to guess the order. And every time you switch
[7:32] one to test if it's the right order, it doubles
[7:34] and doubles and doubles. Okay, so it's kind of
[7:36] been this problem where it's just like, it's
[7:37] impossible to solve because it's constantly doubling
[7:40] and you're never going to get it in the right
[7:41] order because it just keeps getting bigger and
[7:43] bigger. For decades, the smartest people in the
[7:46] field found a whole bunch of really clever ways
[7:48] that made this a little bit faster, but every
[7:50] trick still hit the same wall, and that is that
[7:53] the time required kept exploding as the puzzle
[7:56] grew. About 25 years ago, two computer scientists,
[7:59] Russell Impoglasio and Rohamo Patry, were working
[8:04] with Francis Zane, and they turned that wall
[8:06] into a formal idea, and that's what we call the
[8:08] exponential time hypothesis. This is what they
[8:10] proposed, which basically says that this... explosion
[8:14] is not a failure of human cleverness, but it's
[8:17] just a law of nature. So they're like, look,
[8:19] no algorithm, no matter how smart it is, is ever
[8:21] going to be able to solve a 3SAT without the
[8:24] time blowing up exponentially. Nobody could prove
[8:27] it, but almost everyone believed it. And I think
[8:29] it became one of the foundations that a lot of
[8:31] modern computer science, it used it in a bunch
[8:33] of really weird ways, to be honest. Researchers
[8:36] basically were using this as a measuring stick.
[8:38] So if you could show that solving some other
[8:41] problem quickly could also crack 3SAT quickly,
[8:43] you had shown basically the problem was hopelessly
[8:46] hard. Hundreds of papers, things from scheduling,
[8:49] graphs, logic, a lot more, were all kind of sitting
[8:52] on the same assumption. And so now, fast forward
[8:55] to today, an internal AI model, OpenAI, says
[8:58] that they basically broke it. They said that
[9:01] they found an algorithm that solves the puzzles.
[9:04] a lot faster than the hypothesis allows. And
[9:07] it has checked the proof on Lean, which is the
[9:10] famous computer verification system. So the thing
[9:13] that is the most wild is that of all the people
[9:16] that have kind of, or the people that originally
[9:18] proposed this hypothesis, Impalagi, Zee, and
[9:22] Patry, along with MIT's Ryan Williams, they're
[9:25] basically now leading the public review of whether
[9:28] this has actually proved their own idea. So they
[9:31] came up with this and now they're actually...
[9:32] kind of checking the facts on it so basically
[9:35] what happens next is if the if the disproof is
[9:39] accurate then a huge chunk of a theoretical computer
[9:42] science has to be revisited and i know it's like
[9:45] it's like theoretical computer science people
[9:47] kind of hates these problems sometimes because
[9:49] they're like you know they're like this problem
[9:51] that has no impact on the world is all hypothetical
[9:53] you know but there's actually some interesting
[9:55] things because basically Because they were kind
[9:57] of using it as like a yardstick or a measurement
[9:59] for how hard like a problem was. Every result
[10:02] of previous problems that said, you know, this
[10:04] problem can't be solved faster than assuming
[10:06] ETH, right? They're like, you know, this problem
[10:09] we're currently trying to solve can't be solved
[10:10] any faster than that one. Basically, all of those
[10:14] lose their foundation and researchers are going
[10:15] to have to work out which of those walls that
[10:18] they hit when trying to solve a problem were
[10:19] actually real walls and which of them just simply...
[10:22] they just stopped looking because they thought
[10:24] it would take too long. So I don't think it means
[10:26] that everything becomes easy, but basically the
[10:28] new algorithm is, it's still slow in absolute
[10:31] terms. And I think it's not kind of the famous
[10:33] P versus NP problem being solved, but it means
[10:37] our map of what is possible was wrong in a place
[10:40] that we were very confident about. And if this
[10:43] is the case, then we have to redraw it. So anyways,
[10:45] there's a bunch of interesting implications,
[10:46] but it's always impressive to me to see these
[10:49] new algorithms coming out of models like OpenAI
[10:51] and Anthropic. solving these hard problems, I'm
[10:54] excited by opening eyes, kind of push towards
[10:57] solving some of these math problems, some of
[10:59] these like brain twisters, some of these things
[11:00] that we've thought about for a long time. Because
[11:03] yeah, that's kind of what's exciting to me about
[11:05] science is making progress in these areas. So
[11:08] we'll see what the actual implications are. Maybe
[11:10] we revisit some problems that we previously thought
[11:13] were unsolvable, because it would just take way
[11:15] too long. And we find new ways to solve them.
[11:17] Because if an algorithm could be created for
[11:18] that, maybe an algorithm could be created for
[11:20] some of these other problems as well. Okay, let's
[11:22] talk about what's going on at Verizon. They just
[11:23] signed a $2 billion AI deal with AWS. They're
[11:26] raising their 2026 Outlook. The AWS framework
[11:30] covers distributed compute for agentic and generative
[11:33] AI workloads. Right now, Verizon is saying for
[11:36] a third quarter straight that they believe they're
[11:38] going to be making more money, earnings per share.
[11:40] And so with this, the deal is basically scaling
[11:44] their distributed compute for agentic generative
[11:46] AI workloads. They raised... Their earnings per
[11:49] share guidance to $5 .04 to $5 .06. This is a
[11:54] third straight guidance increase. So Verizon
[11:56] believes they're going to be making more money
[11:57] with this. Their Q revenue this year was $34
[12:01] .4 billion. It's up 1 .5. And their adjusted
[12:04] earnings per share is $1 .31, which is up 8 .3%.
[12:08] So Verizon added $406 ,000. post paid phone subscribers
[12:13] in the quarter, which means you know, this Verizon
[12:15] is growing quickly. And I'm excited to see some
[12:18] of the some of the big bets that they hopefully
[12:21] will take in AI. I've actually interviewed people
[12:23] from Verizon on the show a few years ago from
[12:26] some of their AI labs and some of the projects
[12:28] they were working on. So I'm excited to see them
[12:29] investing more in that and excited to see what
[12:32] those investments ultimately bring. OpenAI has
[12:35] just launched GPT 6 .1. This is their newest
[12:38] model. It's 6 .1 Tera. It's half the price of
[12:41] GPT 6 .1 Sol, which is the midsize model. It
[12:45] now costs a million or $1 per million input tokens,
[12:49] and they're adding computer use and a one for
[12:51] replacing GPT 5 .6 Tera. I actually have personally
[12:55] been using 6 .1 Tera for a bunch of different
[12:57] projects, or sorry, 6 .1 Sol for a bunch of different
[13:01] projects. So if... If, you know, 6 .1 Tera is
[13:05] half the price, I could see that a lot of people
[13:07] would get a lot of really good usage out of this.
[13:10] OpenAI said that 50 .3 % on terminal bench for
[13:15] a max effort is up from 21 % for the previous
[13:18] model. So it basically got twice as good on terminal
[13:21] bench, which is cool. It's going to be inside
[13:23] of chat GPT work and codecs. for plus pro business
[13:27] enterprise and education. It's also going to
[13:29] be on the enterprise and app co -pilot. I mean,
[13:31] obviously they'll roll it out to all of those
[13:32] different places. I'm excited about having it
[13:34] on ChatGPT work if it's, you know, if it costs
[13:36] half as much when you get low on tokens and you
[13:39] need to go to a lower model. I'm always happy
[13:41] when they're able to do that for less. Oracle
[13:43] has just launched 25 Fusion Claw agentic apps
[13:47] across a bunch of different areas. I'm excited
[13:49] to see Oracle making some big bets here. I've
[13:52] heard a lot of rumors about projects that they're
[13:54] leading. Talked to some people there about different
[13:56] things. So I'm excited for the direction that
[13:58] Oracle's going. This project in particular was
[14:01] cool. Fusion Claw is basically Oracle's execution
[14:03] runtime for longer running agent work. And it
[14:07] has continuous replanning and deterministic execution.
[14:10] So it could be halfway through a project. It's
[14:12] like it's OpenClaw, but it's like halfway through
[14:14] a project. And then it will say, look, is the
[14:16] path I'm taking really necessary? Or is it wrong?
[14:20] And then it will replan what it's trying to do.
[14:22] I think that's really cool. Five of the top apps
[14:25] are on the Fusion Cloud CX. So they have service
[14:29] health check and executive briefing documents.
[14:32] They have a bunch of like cool little apps that
[14:33] they've made basically. Oracle Fusion Companion
[14:36] lets users work in Fusion applications through
[14:39] support AI tools like ChatGPT. They didn't disclose
[14:43] the pricing for the applications, but they've
[14:45] created a bunch of new tools. And honestly...
[14:48] whether they're, you know, frontier level or
[14:50] whether they're just good. I'm excited to see
[14:53] Oracle jump in. I mean, I want to see Apple get
[14:55] more aggressive in the space. I want to see Oracle
[14:57] get more aggressive in the space. I want to see
[14:59] Verizon get more aggressive in the space. I want
[15:01] everyone to come up with foundational models
[15:03] and also just tools that build on top of it.
[15:05] I think it's great for the overall industry when
[15:08] you have a lot of competition and a lot of people
[15:10] working on different projects. So yeah, kudos
[15:13] to Oracle. They are obviously raising a ton of
[15:16] money and making a ton of money from selling
[15:18] the hardware and the compute to OpenAI and everybody
[15:21] else. But I think that there's a space for them
[15:24] to do some of the software, some of the algorithms,
[15:26] some of the AI as well. So that's what I'll be
[15:28] excited to see them working on. Thank you so
[15:30] much for tuning into the podcast, guys. Remember,
[15:32] if you want to grow and scale a business or your
[15:34] career with AI, go check out the AI Hustle School
[15:37] community. I have over... 1 ,300 members in there.
[15:40] It is so much fun. If you have questions for
[15:42] me about a business you're working on, projects,
[15:44] what's happening in industry, go join the community.
[15:46] Send me a DM over there. I'll get back to you
[15:48] and I'm excited to see what you build. All right.
[15:50] Catch you all in 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.