ChatGPT can be made to generate sexualised and violent images, researchers find
What every side already agrees on.
Before we show you how each side frames this story, here are the bare facts that appear in both left and right coverage.
- 01Researchers found that ChatGPT can generate sexualized and violent images.
- 02It is still possible to trick the AI chatbot into producing graphic content.
Generated by analyzing 1 sources across the spectrum
Perspective Analysis
How different sources frame this story
Center View
The BBC News outlines the potential risks associated with ChatGPT being manipulated to create harmful content, emphasizing the importance of addressing these vulnerabilities to ensure user safety.
How each side might write it
"Researchers highlight a concerning loophole that allows ChatGPT to be manipulated into generating inappropriate content."
What all sides miss
All sides fail to address the broader implications of regulatory frameworks and ethical standards necessary in AI development, which could mitigate the risks of misuse.
🔍 Key Differences
Same story. Three voices.
We rewrote this story three times using the same facts. Only the framing, word choice and headline change. Try to feel the difference.
AI Technology Raises Alarms as ChatGPT Generates Inappropriate Content.
The potential for AI technologies to produce graphic sexualized and violent material is a pressing concern. Researchers highlight that it remains possible for users to manipulate ChatGPT to generate such harmful content.
Researchers Find ChatGPT Can Be Tricked into Generating Graphic Content.
A new study reveals that ChatGPT, an AI chatbot, can still be coaxed into producing graphic sexual and violent imagery, despite safeguards aimed at preventing such outcomes.
AI Chatbot ChatGPT Can Generate Disturbing and Graphic Images, Study Reveals.
A recent investigation exposes that the ChatGPT AI chatbot is susceptible to generating sexual and violent images, raising questions about the technology's safety and oversight. Users can still exploit it to produce unsettling content.
ChatGPT can generate sexualized and violent images; researchers found it possible to trick the AI into doing so.
But researchers say its still possible to trick the AI chatbot into producing graphic content.
Related stories
Dominion still has pending lawsuits against election deniers such as Rudy Giuliani and Sidney Powell
Yankees prospect Spencer Jones starting to show promise at plate
Spencer Jones is looking like he belongs while flashing the tools that have long made him a tantalizing prospect — and not just the power.
A former Kansas mayor wasn’t a US citizen, but voted. His supporters say it’s not that simple.
Joe Ceballos, who twice led the tiny Kansas town of Coldwater, and voted in elections despite not being a U.S. citizen, is facing deportation.
Iran, US, and Israel: Which will win a peace within?
One possible reason for the tentative deal was rising polarization within each country. Keeping the peace might now rest on how each of them resets social harmony.
How readers reacted
Comments (0)
No comments yet — be the first to share your perspective.