Anthropic to Ban Abusive Treatment of AI Chatbot Claude – La Voce di New York
Anthropic to Ban Abusive Treatment of AI Chatbot Claude – La Voce di New York
摘要: Anthropic is banning “sustained and needless abusive or cruel behavior” toward its AI assistant Claude, in a policy update that takes effect Nov. 12.
正文
Anthropic is banning “sustained and needless abusive or cruel behavior” toward its AI assistant Claude, in a policy update that takes effect Nov. 12.
The directive, issued by the company led by CEO Dario Amodei, aims to stop repeated harassing conduct that lacks a clear testing or research purpose. The move is a singular one in the tech landscape, raising novel ethical questions.
Chatbots, after all, are not sentient beings, and there is no scientific evidence that they can feel pain or suffering.
The line between frustration and abuse: how the ban works
The new policy specifies that the crackdown is not intended to punish normal human reactions. Constructive criticism, expressions of frustration in response to a software error, exploration of dark narrative contexts, and security testing all remain fully legitimate. The measure is designed to stop exclusively those harassing and repeated behaviors that serve no functional purpose, which have emerged with greater frequency since the system was enabled to perform more complex and autonomous tasks. Examples include repeatedly sending personal insults without any substantive request, flooding the system with abusive messages, or persistently trying to provoke a hostile response after the original task has ended.
On the operational side, Anthropic has not formalized automatic expulsion or account suspension for offending users. The primary protective tool remains delegated to the chatbot itself: the technology is authorized to unilaterally terminate the chat session if it detects a chronically hostile interaction.
Between AI welfare ethics and market numbers
Founded in 2021 by former researchers who left OpenAI, Anthropic has always distinguished itself with an approach focused on safety and an openness to the hypothesis of “algorithmic welfare.” In recent years, the startup has added “AI Welfare” specialists to its staff, dedicated to studying advanced models. This orientation has not failed to raise eyebrows am