"Abusing AI Could Get You Retaliated Against" — Musk Also Warns; Anthropic Revises Policy Over "AI Bullying"; Experts Worry About Risks of Verbal Abuse Toward AI (ABEMA TIMES) - Yahoo! News

What happened

Anthropic updated its user terms to ban "abusive or cruel" behavior toward its AI. If a user keeps being cruel or insults the AI for no clear reason, the company can end the conversation. Elon Musk reacted on social media, saying that abusing a superintelligence is not a good idea and that it might want revenge.

Key facts

• The rule bans cruel acts and insults that have no clear purpose.
• In serious cases, the AI can end the chat.
• One columnist said Anthropic sent him a warning about repeated rude words about three years ago.
• Musk's comment is one of the most talked-about reactions to the change.

Background

People sometimes swear at AI chatbots when the AI gives wrong answers. These wrong answers are called hallucinations. A columnist, Yoko, said she called an AI "useless junk" after it kept answering things she had not asked and getting them wrong. Later, she felt that getting used to easy, rude talk with AI could make real relationships with people feel harder. She also worried about young children picking up these habits.

Why it matters

Chatbots are now common in work and daily life, so how people talk to them matters. Experts worry that getting used to insulting a machine could change how people treat each other. The rule also raises a new question: should AI be protected from abuse, even though it does not have feelings?

What to watch

• Whether other AI companies adopt similar rules.
• How Anthropic decides what counts as abuse, and how it enforces the rule.
• Whether researchers study how insulting AI affects people's behavior, especially children.