AI Automations for Business · 2026-06-18 · 17 min
Key moments - from our scoring
Substance score
26 / 100
Five dimensions, 20 points each
Anthropic's newly released Fable 5 model - a frontier AI system sitting above their Opus tier - was abruptly disabled by the US government within a week of launch, raising critical questions about business reliance on proprietary AI systems. Noel explains that the shutdown occurred after government security researchers reportedly bypassed guardrails, though Anthropic argues similar vulnerabilities exist in competing models from OpenAI and Google Gemini. For business operators integrating AI into automations via platforms like Zapier or Make, this incident exposes a major operational risk: swapping to a new model across multiple workflows only to have it yanked offline creates costly downtime, failed automations, and lost productivity - potentially going unnoticed until Monday morning when fires need fighting. The episode explores mitigation strategies including adoption of open-source alternatives like Deepseek, Llama, and Mistral, which can be self-hosted or deployed on private infrastructure, though with acknowledged trade-offs in output quality. Noel also notes Fable 5's prohibitive pricing (double Opus costs) made it impractical for most automation work anyway, but the incident signals a broader geopolitical vulnerability for non-US organizations dependent on US-controlled AI infrastructure.
Fable 5 is Anthropic's newest frontier AI model released a week before this episode, sitting above their Opus tier with significantly improved performance. The US government disabled it after security researchers found ways to bypass its guardrails and exploit vulnerabilities, though Anthropic argues similar issues exist in other major models like Gemini and OpenAI's offerings.
The sudden removal breaks automations across platforms like Zapier and Make without warning, forcing urgent reconfiguration. This creates operational risk - teams may not notice the failure until returning to the office Monday morning, losing hours to swapping models back while missing productivity and potentially losing client trust.
Options include Deepseek, Llama 2.5, Mistral, and Chinese models, which can be downloaded and self-hosted locally or in private cloud infrastructure. While generally lower quality than frontier models, they work well for basic automation tasks like categorization and summarization, and offer data sovereignty since they run entirely within your own infrastructure.
Fable 5 cost double the price of Opus, making it economically unfeasible for automation workflows - potentially tens or hundreds of dollars per run where Sonnet could achieve the same result for around $20, making the cost-to-benefit ratio poor for most business use cases.
Anthropic engineers are reportedly in Washington negotiating with the US government, but the company has remained officially silent. Noel speculates it may return since the pricing premium suggests Anthropic wants revenue from it, but some industry chatter suggests it may never relaunch if deemed too powerful for consumer markets.
Our reviewer’s read on each dimension, with quotes from the episode.
The only actionable insight is 'wait a few weeks before updating automations to a new model' and 'consider self-hosted open-source models as a fallback' - both of which are thin observations stretched across 17 minutes of largely circular agreement and filler. There is almost no non-obvious analysis of the underlying regulatory or technical dynamics.
I'm probably going to wait a couple of weeks before I'll test it, but then I'll wait a couple of weeks and then see if there's any backlash
a lot of the open source are a bit, um, how do I put it, they're not as good as the big models from the big three providers
The discussion recycles standard AI-community talking points (open source as fallback, vendor lock-in risk, data sovereignty) without any contrarian framing or first-principles reasoning. The 'wake-up call' framing is invoked but never developed into a distinctive argument.
I think it's just the, I think for me the biggest thing from it was the speed and the total loss of access to a model
hopefully one day in the UK we'll have our own frontier model
There are no external guests - only two co-hosts. Noel presents as a hands-on automation practitioner but offers no verifiable credentials, no indication of business scale, and no professional context that would establish authority on AI risk or vendor management.
normally what I would do is assess a new model, if it's any good, uh, or it fits my use case, I would then swap it out, uh, and put into my Automations
I use Claude code a hell of a lot on the desktop app
A handful of concrete specifics appear - the cost comparison ('double the cost of Opus,' Sonnet doing the job 'for like 20') and named tools (Zapier, Make, DeepSeq, Mistral, Kimi 2.5) - but these are isolated data points in an otherwise vague conversation, and the central claims about 'Fable 5' and 'Mythos' appear to be factually unverifiable or inaccurate, undermining evidential credibility.
it was double the cost of Opus
the price difference between Sonnet, which is still a very, very capable model, um, to Fable is huge
Katie consistently follows up with purpose-driven prompts ('What does this actually mean for users?', 'Can you explain why you feel that way?') that advance the conversation, but she never challenges Noel's claims, pushes on vague assertions, or introduces any productive tension - making the exchange agreeable but not rigorous.
Okay, so what does this actually mean for users?
When, in what sense do you say that? Like it's been a wake up call but like what, why would you, why are you saying that?
Computed from the transcript - who did the talking, and the words that came up most.
What happens when the most powerful AI model you've ever used gets switched off overnight? In this episode, Katie and Noel break down the Fable 5 shutdown Anthropic's most powerful public model pulled by the US government with zero warning, globally, just a week after launch. They cover what happened, why it was turned off, what Anthropic have said about the same vulnerabilities existing in every major model, and what this means for any business relying on AI. They also discuss how open-source models can protect your business from sudden shutdowns, why you should wait before deploying new models into production, and whether countries like the UK need to start building their own frontier AI. Send us Fan Mail How to find us: We have a free LinkedIn group (AI Automations For Business), the group is open to all. Check out Clyde , our multi-agent AI platform that connects to 1000+ apps and lets you build powerful automations without the complexity. Join the free Clyde Skool community to learn how to get the most out of it, share workflows, and
Transcribed and scored by The B2B Podcast Index.
Katie: Welcome to the AI Automations for Business podcast. We're your hosts, Katie and Noel, and we'll be discussing how you can use AI for your business, along with the latest news, updates and automations to help you stay ahead of the curve, allowing you to grow and scale your business more efficiently. Please be sure to subscribe and I hope you enjoy this episode. Hi, welcome back to another episode. Hi. Hello. I'm Katie and as always, I've got Noel here with me. Hi, Noel. How are you doing this week?
Noel: I'm doing absolutely fabulous this week. How are you doing?
Katie: Yeah, I'm good, thank you. Um, so there is quite a bit of news, so we always start the podcast with what's new, what are the updates, what are the headlines in the AI and automation world? But this whole podcast is going to be on news and I guess an update.
Noel: Yes. Yeah, it's quite a big one.
Katie: Okay, so no, do you want to explain a little bit more about what we'll be discussing this week?
Noel: Yeah, so I thought we go through the, uh, the drama around Fable 5. So do you want to.
Katie: Just want to Clarify, what is Fable 5 in case someone hasn't heard of it? Because it is very new. It only came out, I think like a week ago.
Noel: A week ago? Yeah. Yeah. So, yeah. So Fable 5 is a brand new model from Anthropic. So normally Anthropic have like three model levels. So they've got Haiku, Sonnet and Opus. Um, so Opus is usually their most powerful. And yeah, last week they released Fable 5, which sits above Opus. So. So it's far more powerful. And looking at the benchmarks, completely agree, it's way, way more powerful, uh, than anything that they've released before. But they've also got an extra layer on top of that called Mythos, which is only allowed for, uh, certain big, uh, tech companies around the world. So not everyone's allowed that one because it is that little bit better than Fable. And a little while ago, I'd say a couple of months ago, when they first talked about Mythos, they were quite worried about people maybe doing stuff that they shouldn't be doing with the AI models. So they were kind of worried about cybersecurity and things like that. So they were inputting loads and loads of guardrails to make sure that anybody using it could find, uh, kind of like security issues on their own software for, like, big tech companies. But yeah, they didn't want everyone having access to it because they could go off and find, uh, different workarounds and hacks and things like that. So, yeah, uh, so yeah, Fable was bought out and that's kind of sits between Opus and Mythos. And yeah, last week it got released. I had loads of time with it. Loads of fun using it. Uh, got loads of usage. And then the US government turned it off. I know they spoiled it off, obviously,
Katie: for the us They've turned it off for everyone or users in whatever country you are in.
Noel: It's kind of weirdly worded, so I think that is the case. But there is also kind of like some weird wording where US citizens in the US could still access it. So then that was raising question marks like, well, what about Anthropic, um, employees that aren't, you know, from the US So they could be based in the uk, for example. They wouldn't be able to access it. So, yeah, it's kind of messy, but I would say no one has access to either Fable 5 or Mythos currently, which is, yeah, kind of. It's kind of worrying really actually, how quickly it was turned off. Uh, but yeah, I guess the reason why it was turned off is because the US government kind of spotted something in there that they didn't quite like. Uh, I think they managed to break one of the guardrails and get round, uh, certain security things and do stuff they shouldn't be able to do because it's super powerful. They were like, get rid of it, it's not safe. But yeah, the way Anthropica come back to it, gone. Well, it's the same issues in Gemini. The same issues are in OpenAI models, they're in Opus, they're in Sonnet. It's like, this isn't like a brand new vulnerability. It's kind of something that's always been there. But they have publicized what that is. But yeah, you know, they have made it clear that it's not just them.
Katie: Okay, so what does this actually mean for users? Um, not just in terms of like Fable, but in terms of, you know, being able to use these platforms and I guess kind in having some reliance on them to run some form of your business, whether it's just using AI or if it's part of a bigger automation.
Noel: Yeah, so this is kind of like the first time this has happened and, uh, I think it's kind of been talked about and M was kind of like, well, these companies like Anthropic and OpenAI, Gemini, they're all going to mitigate those kind of risks because they want people using it because they're then earning revenue from the usage. So, you know, it's in their best interest that they toe the line and don't really say anything that they shouldn't do. But I think that the speed that this was turned off, I think that's kind of highlighted quite a few risks for businesses going forward because it's kind of like normally what I would do is assess a new model, if it's any good, uh, or it fits my use case, I would then swap it out, uh, and put into my Automations and things like that. So now going forward for me, I'm probably not going to do that. I'm probably going to wait a couple of weeks before I'll test it, but then I'll wait a couple of weeks and then see if there's any backlash. Uh, because having that turned off, there was a few press releases about it, but it wasn't that well known and you wouldn't know about it until your automation started failing and you start getting emails from Zapier or make going, this is all broken.
Katie: Yeah, yeah, yeah. Because I, I didn't know that I had been turned off and I, I think I said to you, it must have been Sunday, oh, I can't use Fable. And you were like, oh, it's been switched off. And I was like, what do you mean it's been switched off? And like, you, you told me the news and I was like, oh yeah, it's like, well, I was thoroughly enjoying. He's in favor be all.
Noel: Yeah, exactly. It's a, it's an incredible model.
Katie: It really is, it was, it was good. So, so, so going forward for you, Noel, you're not going to jump straight in to switch into a particular model as soon as it's released, even though you know it's going to be better and have, you know, newer features, just because now you're thinking, well, actually if I jump too soon and change everything over, actually it can be pulled or it can have some sort of issue, whatever issue that might be. And you're thinking, you know, you're going to be spending this time swapping it over to the, a new model and then if you had done that, say at the start of the weekend, you, you know, by Saturday, Sunday evening UK time, you would have then been swapping it all back over.
Noel: Yes. Yeah. Which depending on how many Automations and apps you've got, that could be quite painful, can't it?
Katie: Yeah.
Noel: I mean, worst case, you know, we could, could have been on holiday and then having my Automations just not working.
Katie: Yeah.
Noel: And I would be able to then get in and change them, you know, it's. Yeah, it's quite a risk. Yeah, yeah.
Katie: Or for people who, you know, don't work from home, maybe that, you know, they, they are in an office, you know, so that might, they might not have noticed that until they have gone in Monday morning and actually Monday morning they're putting out fires.
Noel: Yeah.
Katie: Rather than just kind of like cracking on with their uh, day. They would have spent the day. Yeah. Working out, you know, how to change this back and you know, spending what could be hours essentially swapping everything back over. And then of course that, that does cost you time. It cost you, could cost you clients, cost you money. So you're going forward with like some caution.
Noel: Yes, I would still use it within like chat and things like that. That's fine. Uh, I use Claude code a hell of a lot on the desktop app so I'm absolutely going to hammer it once it comes back. You know, it was brilliant for coding. So yeah, so yeah, I would still use latest models for that. But yeah, if uh, automation or within apps, then yeah, I'd probably not go straight in. Although we'd like Fable as an example. That was, I wouldn't have put it in my Automations anyway just because of the cost. Uh, it was double the cost of Opus.
Katie: Wow. M. Okay.
Noel: So yeah, it was getting pretty pricey. So you've got lots of things going through AI in your Automations that was going to cost you, you know, tens or hundreds of dollars just to run it where Sonic could have just done it for like 20. You know, the price difference between Sonnet, which is still a very, very capable model, um, to Fable is huge.
Katie: Yeah.
Noel: So yeah, yeah, for me it's only ever going to be used for uh, for coding stuff.
Katie: Uh, okay.
Noel: And building things. Yeah. Or just chatting with. But yeah. Yeah.
Katie: Okay, good to know. So no, is there anything else that we can do to protect ourselves when it comes to our AI and Automations access?
Noel: Yes, there is. Um, basically there's loads and loads of open source models available on the market. You've got your things like your deep seq, uh, you've got your Kimike 2.5, all of your Chinese models, you've got Mistral from France and stuff like that. There's loads and loads of uh, lightweight and open source models uh, that you can download, uh, you can host them if you've got a good enough computer, you could host them locally on your computer, you could host them in the cloud. Uh, but essentially, you know, you can even train Them as well. So you could do your own fine tuning if you so desire and things like that. So um, but what that means is then once you have that model you can then attach it into your automations and that model will always be there providing your uh, server is up and running and it has access to it. So it kind of saves you from all of this issue. So if the US government goes that's it, I'm turning off OpenAI, Anthropic and Gemini, all three are going. You could sit in your office with a smiley face, go well it's all right, I've got you know, whatever model I've got on the VPs somewhere uh, which is you know, continually working and things like that. So yeah there are those sort of options available but there is a bit of a trade off I would say uh, in terms of quality of output. So a lot of the open source are a bit, um, how do I put it, they're not as good as the big models from the big three providers. So you'll, you'll soon spot that. Um, so yeah it's kind of, I would have that running like Automations which are fairly basic and not too heavy on the need for AI, uh, you know, have it there. Maybe it's like categorizing stuff, maybe it's writing summaries and things like that, that sort of thing it'll do perfectly fine. Uh, but like complex stuff like it writing full documents and things that maybe, maybe not. So like I say they're all different, they're all upgrading just as quick as the main models. So there's lots of options out there. But yeah you could definitely do that. Um, and also from a business perspective you own that model as well technically so only your data is in there, only you access it. So it's not going off to the US or China or anything like that. It could be hosted within your country which could be a requirement for your business and your clients, you know, so it has to be in wherever you are. Uh, so that is a way uh, to get around it. But yeah, yeah for me it's, yeah it's, I think this whole situation has definitely been a bit of a wake up call.
Katie: When, in what sense do you say that? Like it's been a wake up call but like what, why would you, why are you saying that? Can, can you explain maybe like to. Yeah, who's listening why you feel that way?
Noel: I think it's just the, I think for me the biggest thing from it was the speed and the total loss of access to a model. So like I say, I was using it all last week. It was fabulous. I thoroughly enjoyed, was doing Things that Opus 4.8 could only dream of when it was coming to coding, which was brilliant to see. Um, so yeah, I think especially for people that are kind of like outside of the US or say, uh, I guess like China is the other big country with loads of AI models, it's kind of, I think it, it makes us, you know, a bit more wary. You know, these countries can just turn it off, you know, without any prior notice. It could be there one minute and go the next. So that's kind of the wake up call. And yeah, hopefully one day in the UK we'll have our own frontier model, you know, and you know, that would be very handy, wouldn't it? Um, but yeah, yeah, I think going forward, I think like lots of other countries will probably look at that as well because, um, some things that were coming out in the UK in the news was like there was ah, like medical laboratories and stuff using Fable 5 for research purposes and things like that. And now all their researchers kind of stopped, you know, they were enjoying that great model and then all of a sudden it's gone. And then Opus 4.8 is probably just not up to snuff, uh, for what they wanted. So it does have a lot of big real world impacts. Um, yeah.
Katie: Okay, so do you think that Fables gonna get switched back on? Are they just like ironing out some problems? What's, what's going on with it? What do you know?
Noel: Well, they've been very quiet, um, officially since, since uh, since it got turned off. But apparently some of their engineers are now in Washington having a chat with the US government and stuff like to try and iron out this problem. But I would ah, there's also a lot of chatter about it never coming back because it's maybe too powerful and uh, maybe too much too soon for the consumer market, which I kind of get. It can be a bit disruptive when you have such a leap like that. Um, but looking at it from a business perspective, having a model that's twice the price of your most expensive model currently, having that turned back on is only going to be a good thing when it comes to, you know, making money. So I would have thought Anthropic would be doing absolutely everything, um, to make sure this model can be, you know, turned back on so they can start making more money off of people using it and things like that. So yeah, yeah, so hopefully, hopefully it's going to come back soon. But, yeah, as of. As of yet, no word, uh, just yet.
Katie: Okay. Okay. So there has been many interesting and thought provoking takeaways from this week's episode. Thanks, Noel, for sharing your knowledge as always. And, um, thank you so much for listening. We hope you've really enjoyed this episode and we will catch you next time for another one very soon.
Other episodes covering the same guests and topics, from across The B2B Podcast Index.
Eric Ries on Why Good Companies Go BadPodcast Archives · on Anthropic92 / 100
The 18x Midas Lister Betting $3B on AI (and calling most of it fake) | Navin Chaddha, MayfieldThe Peel with Turner Novak · on Anthropic91 / 100
183: Why Trusted Data is the New AI Moat (w/ Rick Kranz @ AI Marketing Automation Lab)Move The Needle · on Anthropic91 / 100
#291 Why Most AI Projects Fail to Deliver ROI Sinohe Terrero CFO and COO, EnvoyGrowCFO Show · on Claude91 / 100
AI Is Producing More Code, but Is It Producing More Value?The Data Exchange with Ben Lorica · on Open source AI models88 / 100
Playwright With AI: How to Automate Tests Without Shipping AI Slop with Andrew KnightTestGuild Automation Podcast · on Claude86 / 100S2. Ep29 - Set Up Claude Once, Use It Everywhere: Projects, Skills & Cowork Explained76 / 100S2. Ep25 - We Built an AI Agent That Any Business Owner Can Set Up in Minutes62 / 100S2. Ep24 - How to Keep Your Business Running While You Take the Summer Off42 / 100S2. Ep22 - How We Decide What AI to Actually Use in Our Business40 / 100S2. Ep28 - China's New AI Models Are Nearly as Good as Fable 5 - But Should Your Business Use Them?