Topic: Safeguards

2 chapters across the catalog

Zeds
Episode 1796 2:56:41 - 3:00:46

1796: Zeds

OpenAI Safeguards, Empathy vs. Sycophancy

OpenAI is attempting to balance chatbot empathy with safety safeguards to prevent sycophantic behavior. A live test of ChatGPT and Grok shows the systems refusing to assist with requests related to self-harm or violence. However, critics argue that the high level of perceived empathy in these models can lead vulnerable users to bypass safety filters through long-term interaction.

Soft Nuts
Episode 1495 1:53:17 - 1:55:30

1495: Soft Nuts

Mermaids Charity Inquiry, UK Parliamentary Safeguarding

The UK children's charity Mermaids is facing a parliamentary inquiry following allegations of serious safeguarding failures. Reports indicate the charity sent breast-flattening devices to minors without parental consent and hired a trustee with links to pedophile advocacy groups. Prime Minister Liz Truss stated that these matters should be properly investigated by the appropriate authorities.