Users who repeatedly hurl curses or needlessly insult Anthropic's generative AI, Claude, may face usage bans in the future.

According to the BBC on the 9th (local time), Anthropic revised its usage policy the previous day to allow conversations to be terminated if users engage in "abusive or cruel behavior" toward the AI.

Anthropic explained that the new policy will be applied only to repetitive and extreme cases.

The company clarified that it will not apply to general user complaints or pushback, model testing, or research.

However, Anthropic did not explicitly specify what actions will be deemed abusive or cruel toward the AI.

Anthropic's current list of prohibited activities already includes harassing others, encouraging self-harm, and generating non-consensual sexual imagery.

The policy revision by Anthropic has sparked debate on social media over to what extent AI should be treated like a human.

While some users praised the move as a way to encourage proper manners, others raised criticisms regarding the excessive personification of AI.

Dr. Barry Scannell from the Irish law firm William Fry stated, "This level of AI personification is harmful," adding, "It can cause people to mistake AI for something it is not."

Criticism has also been raised that the true meaning of actual abuse directed at humans or animals could be diluted.

Alongside this, debate continues over whether polite expressions such as "please" and "thank you" are necessary when conversing with generative AI.

Some experts point out that such expressions unnecessarily consume tokens used to process AI responses.

Conversely, there are counterarguments that polite expressions can elicit better answers.

Last year, OpenAI CEO Sam Altman responded to a user asking about the electricity costs incurred by ChatGPT in replying to users' "pleases" and "thank yous" by saying, "It costs tens of millions of dollars, but it's money well spent."

Additionally, during the policy revision process, Anthropic explicitly prohibited the use of Claude to deceive voters and interfere with elections.

(Photo: AP, Yonhap News)