Jul 7, 2026 · 43 min · 7 segments
With the weather warming up again, Rob and Lawrence dive into the seasonal world of BBQ tech, and explore whether you should go charcoal, gas... or even electric. Plus the latest technology news.
Rob ChipperfieldHost
Lawrence MicallefHost
But first, this week's technology news, and starting alphabetically with AI, and we have the return of the Fable.
There, there was this sort of interesting thing where they, you know, they, they sort of provide potentially the, the-- a way of, uh, discovering security vulnerabilities, and if that's available to the bad guys, in inverted commas, um, then that might, you know, cause these to be discovered by the people you don't want them to, rather than giving chance for the people who do want to, uh, you know, who do want to find these things to be able to fix stuff first.
Mythos was where it started, and then they said, "Oh, no, this is far too dangerous.
Um, we'll, we'll trim it back." Um, and there's this sort of weird thing of if you try and use Fable for anything too security-related, it bounces you back down to Opus, which was the previous sort of, uh, best model out there.
Um, I, I think this was, this was, yeah, a fantastic marketing opportunity, whether, uh, deliberately or otherwise by Anthropic of like, "Oh my gosh, our things are too good to be, uh, to, to be, you know, in, in public." Uh, and, "Oh, look, now you can get your hands on them." Nonetheless, let's, uh, let's see how it goes this time.
Again, Mythos still restricted to a select group of US-based organizations approved by the federal government.
And, and, yeah, my, my own sort of use of this in, in the day job, um, yeah, prior to the-- these models becoming, uh, available again, is that, you know, it is pretty good at finding security problems if you've got the source code of an application available.
Um, uh, and of course, that does give a potential opportunity for, for companies to test their own code, and in fact, the, the terms of use say you can only test your own stuff.
Um, but of course, the source code for an awful lot of open source stuff is out there.
Open source maintainers doing it often voluntarily aren't going to have the thousands or tens of thousands of, of pounds of budget, uh, to, to be able to spend on these models.
And so yeah, there is, there is the risk that the bad guys are gonna get there first, perhaps better funded than the volunteers.
But first, this week's technology news, and starting alphabetically with AI, and we have the return of the Fable.
There, there was this sort of interesting thing where they, you know, they, they sort of provide potentially the, the-- a way of, uh, discovering security vulnerabilities, and if that's available to the bad guys, in inverted commas, um, then that might, you know, cause these to be discovered by the people you don't want them to, rather than giving chance for the people who do want to, uh, you know, who do want to find these things to be able to fix stuff first.
Mythos was where it started, and then they said, "Oh, no, this is far too dangerous.
Um, we'll, we'll trim it back." Um, and there's this sort of weird thing of if you try and use Fable for anything too security-related, it bounces you back down to Opus, which was the previous sort of, uh, best model out there.
Um, I, I think this was, this was, yeah, a fantastic marketing opportunity, whether, uh, deliberately or otherwise by Anthropic of like, "Oh my gosh, our things are too good to be, uh, to, to be, you know, in, in public." Uh, and, "Oh, look, now you can get your hands on them." Nonetheless, let's, uh, let's see how it goes this time.
Again, Mythos still restricted to a select group of US-based organizations approved by the federal government.
And, and, yeah, my, my own sort of use of this in, in the day job, um, yeah, prior to the-- these models becoming, uh, available again, is that, you know, it is pretty good at finding security problems if you've got the source code of an application available.
Um, uh, and of course, that does give a potential opportunity for, for companies to test their own code, and in fact, the, the terms of use say you can only test your own stuff.
Um, but of course, the source code for an awful lot of open source stuff is out there.
Open source maintainers doing it often voluntarily aren't going to have the thousands or tens of thousands of, of pounds of budget, uh, to, to be able to spend on these models.
And so yeah, there is, there is the risk that the bad guys are gonna get there first, perhaps better funded than the volunteers.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.
