We held back on posting about ChatGPT, the new artificial intelligence platform by OpenAI, due to the internet being awash with information and opinion about it. After all it is only a few month old and much of it is still under development.
Nonetheless, we thought a new article in ArsTechnica titled “OpenAI checked to see whether GPT-4 could take over the world” to be worthy of mention. It highlights some of the potential and emergent capabilities of this new generation of AI which may make it somewhat risky for society in general. A key experiment that was conducted by researchers on the new version of ChatGPT-4 highlighted that it autonomously had the capability to dupe or ‘socially engineer’ humans in order to fulfil its objective. This information seems to reinforce the concern that many experts in the cybersecurity world have about the AI being used as force multiplier for cyber attacks.