OpenAI, Anthropic aim to protect teens from AI chatbots—here’s how

Watch OpenAI, Anthropic aim to protect teens from AI chatbots—here’s how - Rizzle on Dailymotion

Transcript

00:00OpenAI, Anthropic aimed to protect teens from AI chatbots, here's how.

00:05OpenAI and Anthropic announced new efforts to spot underage users, as pressure mounts

00:10on AI companies to keep teens safe and lawmakers sharpen their knives.

00:15OpenAI says it's updating the rulebook that governs how ChatGPT behaves, especially when

00:20chatting with users aged 13 to 17.

00:23The chatbot's updated model spec now adds four new principles designed to put teen safety

00:28first, even if that means dialing back some of the chatbot's usual enthusiasm for unfiltered

00:33intellectual exploration.

00:35The new guidelines tell ChatGPT to gently steer teens toward safer options when conversations

00:41drift into risky territory, encourage real-world relationships and offline support, and, perhaps

00:46most notably, treat teens like teens.

00:50Lawmakers have been increasingly worried about the impact of AI chatbots on mental health,

00:55particularly for younger users.

00:58OpenAI is facing a lawsuit alleging that ChatGPT shared harmful guidance with a teenager

01:03who later died, raising concerns about AI safety.

01:07In response, the company rolled out parental controls and barred suicide-related discussions

01:11with teens entirely.

01:13Now, OpenAI says ChatGPT will push users toward trusted adults, emergency services, or crisis

01:20resources when it detects signs of imminent risk.

01:23If the system thinks you're under 18, teen safety features kick in automatically, though

01:28adults who get mistakenly flagged can verify their age.

01:32Anthropic is taking a stricter approach.

01:35It doesn't allow under 18 users at all, and it's building tools to detect and boot

01:38minors from its chatbot, Claude.

01:41Anthropic also shared details about training Claude to avoid reinforcing harmful thoughts,

01:46including around suicide and self-harm.

01:49Its newest models are reportedly less sycophantic in their AI speak, with the Haiku 4.5 model correcting

01:55that behavior 37% of the time.

02:07To be continued...

02:11...

02:24...

02:27...

02:29...