Emma Gillespie

over disturbing content, including descriptions of gun violence.

So staff at OpenAI actually wanted to report this to authorities.

The account ended up being banned, but they did not alert local police about these concerns.

Well, OpenAI has said basically that this account didn't meet its threshold for what it defines as a credible or imminent plan for serious physical harm to others.

So they certainly assessed the nature of these prompts and conversations, but deemed that it just didn't meet that threshold.

But Altman since has apologised publicly.

He said quote, I am deeply sorry that we did not alert law enforcement to the account that was banned in June.

Families of victims though have not accepted that apology.

They have accused him of playing a game with our community where we were the only people who could ever lose.

So there are seven families who have now brought ongoing legal action against OpenAI.

In Canada, against a US company.

So that's all separate legal action, but I think it paints a really important picture about these concerns around potential harms involving chat GPT.

Yeah, well, I guess the short answer is it,

depends platform to platform.

There are no real regulations here.

And many advocates for AI protections in this space have really noted their concerns around the fact that it's kind of the Wild West out there still.

This technology has evolved so rapidly in recent years that those catchments and regulations aren't

shored up yet, and of course vary between platforms and jurisdictions.

But I did think that this piece of research out of the UK was quite interesting this year.

The Center for Countering Digital Hate tested 10 leading AI chatbots, and it found that 80% regularly assisted users seeking help with violent attacks.