Hi - we are seeing the evolution of many GPT engines that do not have the ethical boundaries of more main stream these for instance are being trained on malware code and phishing emails to craft higher quality targeted attacks

26 September, 2024 12:12

Do you see a future where GPT engine providers will be forced to legislate the use of their technology (perhaps via an ethics charter or suchlike) or is it already too late and the genie is out the bottle?

I think, personally it's the later - as there's so much freely available AI engine tech already out there in the wild to use. But perhaps the later generations of ChatGPT and similar tools can be distributed with more safety in their design, depending on if there's external pressure to do so. It's a thought anyway.

26 September, 2024 12:27

as you say genie may be out of the bottle - on one of the other questions I think a comment was made about arms race - and I think that we will see this for a while

and where controls are implemented/mandated - the likelihood is that attackers will look to circumvent them

26 September, 2024 12:43

Thanks Rich, there's also the point from Paul and legislation will always be reactive / after the point, so yes the arms race point is well made

26 September, 2024 12:46