Anthropic’s new AI rules explained

Summary
In this news segment from LiveNOW from FOX, anchor Bowen Kedrowicz interviews Satya Thallam, Senior Fellow at the Foundation for American Innovation, regarding Anthropic’s updated usage policy that bans “sustained and needless abuse or cruel behavior” toward its Claude AI models. Thallam explains that while the rule uses anthropomorphic language, it functions as a terms-of-service boundary to govern non-deterministic LLM behavior, manage product liability, and mitigate risks of jailbreaks and harmful outputs.

What is shown

• [00:00] Live broadcast opening with nighttime skyline views, Fox news lower thirds, and an interactive viewer poll asking: “Do you think AI companies are doing enough to keep AI safe?”
• [00:58] Split-screen interview format featuring host Bowen Kedrowicz in studio and guest Satya Thallam via remote video link.
• [04:09] B-roll footage showing close-up shots of server racks, flashing networking switch LEDs, ethernet cabling, and a smartphone screen displaying the Anthropic logo.
• [04:29] B-roll showing code interfaces, terminal windows with hex data and assembly-style code, and a text generation prompt interface creating birthday party ideas.

Claims & numbers

• The presenter states that Anthropic updated its usage policy to prohibit “sustained and needless abuse or cruel behavior toward models,” enabling Claude chatbots to refuse or end interactions in extreme cases where users act cruelly with no discernible purpose [00:16].
• The presenter notes that Anthropic also restricts Claude from responding to requests involving child sexual abuse material (CSAM) or instructions facilitating large-scale violence or acts of terror [08:18].
• Thallam claims Anthropic is “possibly going to have the largest IPO in history” potentially within a matter of weeks or by year’s end [05:51].
• An on-screen LiveNOW poll displays viewer responses: 48% No, 8% Somewhat, 27% Yes, and 16% Not sure [14:48].
• Thallam references recent letters and warnings from the Federal Trade Commission (FTC) chair cautioning AI developers that anthropomorphizing models does not absolve them of legal liability for their systems [16:58].

Notable quotes

• [01:50] “And it’s it’s just a description of the terms of service.” — Satya Thallam
• [16:16] “The idea of agency, of a living being, of another person means that they’re making their own choices. ‘I didn’t make them for them, it made it for itself.’” — Satya Thallam
• [17:13] “If you start to think that a thing has its own beliefs and and preferences, then what responsibility do you have?” — Satya Thallam

Assessment
This is a standard live cable news interview and policy analysis segment, not a product demo or launch event. The discussion provides contextual analysis of Anthropic’s updated acceptable use policy and broader legal and regulatory considerations without demonstrating live interactions with Claude.

Described by gemini-3.8-flash on 2026-10-16 from the video’s audio and frames.