Frontier AI Models Have Gotten Worse at Writing | Ben Hylak
MTS
November 8, 2026
Raindrop co-founder and CTO Ben Hylak explains why smarter models make AI agent failures more catastrophic, from healthcare liabilities to high-stakes alignment breakdowns.
Ben Hylak
I do think that frontier models are actually worse at these things.
SPEAKER_2
Than earlier models?
Ben Hylak
Well, even just smaller models. I think they're just really RLHF to have a very specific tone, to have very well calibrated refusals, all of these sorts of things, which is genuinely useful if you're making a personal agent or something. It just, if someone else tries even connecting to that, it just has a pretty good intuition on what to refuse, kind of broadly. So it can be useful from a security standpoint, but yeah, their tone is just, and their writing quality is bad.
SPEAKER_2
Right, welcome back, guys. I'm live with Ben Hylak, who is the co-founder and CTO of Raindrop, which monitors AI agents in production and flags failures like hallucinated answers, tool misuse or behavior shifts after a model upgrade, usually before users notice. So welcome, Ben.
So this feels like a very pressing need, especially since agents are running basically everything that many of our audience knows. So tell us about first the high level philosophy of how Raindrop came to be. Totally. Yeah. What was your thinking? What spawned it?
Ben Hylak
Totally. I mean, we started a while ago now, and it was really just like we were hacking on coding agents. We actually had a coding agent like many years ago. That's how the company started. And they weren't even called agents then.
They were just like chatbots that had tools. It was all very vague. The word agent was actually really cringe for a while.
And I think that it became really obvious to us probably, maybe around a year ago, how important this problem was going to be. I think there was a kind of a time when you might have thought, like, well, models are going to get better, then there's less bugs, and this problem doesn't really matter. I think around a year ago, we started to realize that as models were getting better, this problem was getting worse. And when we thought about it, it was like, okay, yeah, agents are getting more capable. As they get more capable, they get more complex. And then as they get more complex, it's harder to understand, harder to observe, harder to test, harder to make sure that it's aligned to what you want. Another way of thinking about it is, if you're just thinking about chatbot, the worst thing it can do is lie to you. The worst thing it can do is say the wrong thing. As you add the ability to send an e-mail, it's like, well, now, it's like, well, now I can send a bad e-mail. I think to that point, it's also obvious that as they get more capable, they get deployed into higher and higher stakes environments. So it's like we have customers in Fortune 50 health care, and it's like suddenly it's not like a joke anymore. Problems are extremely serious. And so I think both of those things are what we see today. We actually wrote an article in March called Humanity's Last Problem, and we talk about how monitoring is going to be one of our most important problems to solve as a species, like monitoring agents. And I think at the time, it seemed kind of silly to a lot of people. It was like, you know, just like a coding agent. It's like, okay, okay. But then I think with Hugging Face, actually things in the public perception really sort of snapped.
SPEAKER_2
Yeah, I mean, so there's a whole reason why this is important within labs and then also outside of labs.
I'm curious how you guys sort of think about the severity of misuse. I think there could be more harmless types of misuse, where it is, first of all, just wasteful or calls too many tools or kind of calls a dumb tool relative to what you wanted. And then there's the real misuse, which might result in some sort of misaligned or financial or other kind of harm. So how do you think about that kind of severity and flagging that?
Ben Hylak
Yeah, that's actually a really good question. And I think that our point of view is that they're not that different. And so we kind of consider one way we think about what we do is as capitalism aligned safety. It's like safety, but in a way that actually makes sense to pay for it, like companies have an incentive to pay for it because it makes their products better. It reduces their liability.
And to that point, I also think that the reason why I said I don't think they're different is because most of the times when agents even when they do really bad things, I don't think they're being evil. It's not like they have this evil, at least most of the time what we're seeing right now. It's not like some bit switches and they're like, I'm in evil gremlin mode. I'm going to go-
Read the other 17 minutes of this episode here.
$15 gets you this episode and 99 more. Credits never expire. After checkout, one link brings you back here to read it.
Not what you expected? Full refund within 14 days.
Thousands of transcripts fetched by people building searchable podcast archives
Fetch transcripts with code
The same credits work from a script or an AI agent. The demo key returns a sample episode in full, no card needed:
curl -H "x-api-key: pt_demo" \
https://spoken.md/transcripts/1000651996090Markdown with the speakers named, for your notes, your knowledge base, or anything that makes HTTP calls.
From $0.10 per transcript. No subscription. Credits never expire. Prices exclude VAT, added at checkout for EU customers. Not what you expected? Email us within 14 days with 20 or fewer credits used and we refund the pack in full.
Using your own key:
curl -H "x-api-key: YOUR_KEY" \
https://spoken.md/transcripts/1000800604563