So there are volume problems, that's obvious, but there are uncertainty on the code that is produced.

Yeah, I think that's correct.

I think the provenance piece I mentioned at the top of the podcast is new.

You say that, well, maybe it's not a new problem, it's just the velocity.

Well, that's also true in terms of automobile impact.

And automobiles have impacted from the very first day of their inception.

How are the differences now? They do it at several hundreds of kilometers an hour rather than at a...

What else do you think should be implemented? even if possible, automated and added by humans? What is your perspective right now?

So there are volume problems, that's obvious, but there are uncertainty on the code that is produced.

Yeah, I think that's correct.

I think the provenance piece I mentioned at the top of the podcast is new.

You say that, well, maybe it's not a new problem.

It's just the velocity.

Well, that's also true in terms of automobile impact.

And automobiles have impacted from the very first day of their inception.

What else do you think should be implemented? even if possible, automated and added by humans? What is your perspective right now?

So if you throw in more than one agent, more than one agent model, maybe even more than one agent vendor, you would expect better results, right?

I always think so.

And I think the, again, you're quite right that we mustn't over sensationalize this.

And I think one of the pitfalls is attributing human emotion or human intent.

Exactly.

And I have this battle with many others in the AI space who say, well, we can understand intent and I don't believe that we understand the human's intent and possibly not even our own.

So how can we possibly understand the intent of AI? That's a separate conversation.

So I

What do you expect what we can do as, as vendors, as standards, as analysts, as customers to, to leverage the effects of all this agent communication and collaboration while preventing everything that went wrong during these experiments? Long question, sorry.

Well, that's a...

No, it's a brilliant question, and, uh, let me try and, uh, be equally brilliant in my turn.

So I think the first, I think the first chunk of this is that we must see every challenge of instruction as a part of a larger context, and we must make sure that that larger context is explicitly included in directions.

So right now, we have to say, um, "Please refer to KuppingerCole's security policy and use that as a guide to correct behavior," for example.

And the larger piece of this is actually emerging as a way of instructing these things.

So rather than, as you say, going back to the, well, I'd say the early stages, we're talking about, you know, perhaps a year ago, if that, we would put a prompt into a agent, and it would give a response.

Weaken

Jonathan, your opinion on that?

Thank you.

So I've been thinking about this a lot, as I'm sure everybody has.

I think there is a false equivalence here dressed up as a warning.

I think OpenAI deliberately disabled its guardrails for a control capability test, which is something that happens in, in testing stages.

And I think that a team that's running a coding agent in CI has not switched anything off and isn't stress testing and punting model.

So we've got a, an equa-- a, an equivalence between a lab experiment with ordinary enterprise use to manufacture alarm.

So how can we prepare for such a situation, either from a technology perspective, which I don't think is the actual case, but from a process, from a design, from an identity managed perspective, where, where should we start? What, what are takeaways that we really can, can take from that once the, uh, yeah, the, the, um, the, the Skynet scenario has been stripped off the, the event? What can we really do to prevent this? Jonathan, maybe you want to start.

Jonathan, your opinion on that?

Thank you.

So I've been thinking about this a lot, as I'm sure everybody has.

I think there is a false equivalence here dressed up as a warning.

I think OpenAI deliberately disabled its guardrails for a controlled capability test, which is something that happens in testing stages.

I think that a team that's running a coding agent in CI has not switched anything off and isn't test testing and printing model.

So we've got an equivalence between a lab experiment with ordinary enterprise use to manufacturer law.

Where should we start? What are takeaways that we really can take from that? Once the Skynet scenario has been stripped off the event, what can we really do to prevent this? Jonathan, maybe you want to start?