Snapchat users left panicked after My AI goes rogue with ‘ghost’ picture

Snapchat users have been spooked after its AI chatbot seemingly went rogue and posted an image to its story.

Before the surprise post users did not even realise the chatbot, called My AI, could post its own content to its story.

The picture showed two interlocking colours, which some users interpreted as a picture of a wall and a ceiling – prompting concerns the bot was taking its own photos with users’ cameras without their permission.

The post was quickly deleted, and My AI went offline for a short time. When it came back online, users were able to ask it what had happened.

My AI responded that the post was a ‘fun way to mix things up’ and apologised if it had ’caused any concern’.

It also said that the post was a ‘spooky ghost prank’ that had ‘given us quite a scare’.

Snapchat acknowledged that issue, saying that My AI experienced a ‘temporary outage’ that’s now been resolved.

While it’s unlikely that the chatbot was pranking users, the likely explanation for its responses saying so are ‘hallucinations’.

AI hallucinations are false information or facts that have been observed in large language models (LLMs) like OpenAI’s GPT-4 and Google’s Bard.

They are not based on real data or events, but are often presented with confidence and authority. It was the chatbot’s way of explaining the glitch to users, inventing stories in response to questions, rather than giving factual information.

Earlier this year, the social media company introduced My AI as ‘an experimental, friendly, chatbot’ designed to be a ‘personal sidekick’ to Snapchat users.

Readers Also Like: New York Times sues Microsoft, ChatGPT maker OpenAI over copyright infringement

The feature powered by the viral AI chatbot ChatGPT is a technology still under development, so issues like these are to be expected.

As for users worried about My AI taking pictures without your permission, as with any app, users allow access to certain information when they download an app.