Skip to content

TechnologyDeveloping

Anthropic updates Claude usage policy to bar sustained abusive or cruel behavior toward its models

By

Coverage spectrum

Covered by 9 outlets across the spectrum.

11 articles in total, 1 of them opinion. The spectrum counts each outlet once and leaves opinion out. How we measure bias

Laptop screen showing Anthropic website with ANTHROPIC wordmark and Claude logo and branding
Image via The Washington TimesHow images are chosenReport this image

Anthropic, the company behind the Claude AI chatbot, updated its usage policy on Thursday to prohibit “sustained and needless abusive or cruel behavior” toward its AI models. The company said the rule applies only in “extreme cases where users repeatedly act cruelly toward our models, with no discernible purpose,” and does not cover ordinary frustration, pushback, dark creative themes, or model testing and research. Reports say the change takes effect November 12. Cutting off a conversation remains the main enforcement tool, and Anthropic did not immediately say how it would distinguish frustration from cruelty. The update comes amid a wider debate over whether AI systems can be conscious. Anthropic has said it is highly uncertain about the potential moral status of Claude and other models.

Headline and summary written by Mantle’s AI from the coverage below, under published rules. How this works

What happened

Key points

Each point shows who reports it, from Far Left to Far Right.

  1. Anthropic updated its usage policy on Thursday to prohibit “sustained and needless abusive or cruel behavior” toward its Claude models.

    Reported byFar Left1Left2Center1Right2Far Right2See the outletsHide the outlets
    Far Left
    The Daily Beast
    Left
    Business Insider, CBS News
    Center
    BBC News
    Right
    New York Post, The Washington Times
    Far Right
    The Daily Caller, Breitbart News

  2. Anthropic said the rule applies only in extreme cases where users repeatedly act cruelly with no discernible purpose, and does not cover common frustration, pushback, dark creative themes, or model testing and research.

    Reported byFar Left1Left1Far Right1See the outletsHide the outlets
    Far Left
    The Daily Beast
    Left
    CBS News
    Far Right
    Breitbart News

  3. Anthropic did not say how it would tell frustration from cruelty; ending abusive conversations remains its main enforcement tool.

    Reported byFar Left1Left1Far Right1See the outletsHide the outlets
    Far Left
    The Daily Beast
    Left
    CBS News
    Far Right
    Breitbart News

  4. The policy change takes effect on November 12.

    One side only: the leftReported byFar Left1Left1See the outletsHide the outlets
    Far Left
    The Daily Beast
    Left
    Business Insider

  5. The update is Anthropic’s first policy rewrite in over a year and also adds restrictions on election interference, weapons development and surveillance.

    One outletBreitbart News· Far Right

What is contested

Where accounts differ

Points the coverage does not agree on, and who says what.

  • Whether the rule is a serious policy on how AI models are treated or a trivial matter

    Right outlets, especially the Daily Caller opinion column, mock it as a distraction from AI risk; Breitbart and the Washington Times describe it as a protection against abuse of AI; the Daily Beast and CBS report it as a rule on user conduct without judging its importance.

  • Whether Claude can be harmed, and so whether it needs protection from cruelty

    Anthropic says it is highly uncertain about the moral status of Claude and other models; the Washington Times presents the policy as protecting AI, while the Daily Caller opinion column says the company has moved on from warnings about AI risk to chatbot feelings.

  • How the rule will be enforced in practice

    Anthropic says the rule covers only extreme cases with no discernible purpose; the Daily Beast says it is not clear how frustration will be told apart from cruelty, and CBS reports the company did not respond to a question on what counts as abusive.

How each side tells it

Compare views

  • the left

    3 outlets

    Left headlines treat the change as a company rule about user behavior: the Daily Beast, the most pointed of them, calls it a ban on users who bully the chatbot, while CBS uses a playful Elvis reference and Business Insider frames it around commentators being split.

    Why am I so mean to my AI?
    Business InsiderLeftAll left-leaning headlines

  • the center

    1 outlet

    The single center article, from BBC, headlines the change as a ban on users being ‘cruel’ to the firm’s AI systems and builds its description around the company’s own wording.

    Anthropic bans users from being 'cruel' to its AI systems
    BBC NewsCenterAll center headlines

  • the right

    5 outlets

    Right headlines mostly treat the rule either as a joke about being nice to chatbots or as a protection for AI, with the far-right Daily Caller most mocking, and its opinion column calls it a distraction from AI risk; Breitbart’s description leads with the election, weapons and surveillance rules.

    Anthropic Is Banning 'Cruel Behavior' Toward Claude. Claude Is Still Not a Person.
    ReasonRightAll right-leaning headlines

The words they chose

Language to notice

Words that carry a judgement beyond the facts. Tap a highlighted word anywhere on the page to see why.

  • Left3

    critics saidpeople are sayingBully

  • Center0

    No loaded words in its own voice.

  • Right8

    strangeoddball corners of Silicon Valleyinsistwelfare

In the outlets’ own words

  • Sensational language

    strangeMore neutral: The update

    Calling the update strange is the outlet’s own judgement rather than a description of the policy change.

    The strange update came as debate continues to rage about regulating AI, and as oddball corners…

    Used byNew York Post (Right)

  • Loaded modifier

    oddball corners of Silicon ValleyMore neutral: some people in Silicon Valley

    Labels the people who hold a view as oddballs before the view is weighed.

    …update came as debate continues to rage about regulating AI, and as oddball corners of Silicon Valley insist the tech has a soul.

    Used byNew York Post (Right)

  • Loaded verb

    insistMore neutral: believe

    Insist signals the outlet doubts that the technology has a soul, when it could simply report the view.

    …continues to rage about regulating AI, and as oddball corners of Silicon Valley insist the tech has a soul.

    Used byNew York Post (Right)

  • Delegitimizing label

    welfareMore neutral: treatment of models

    Placing the word in scare quotes dismisses the concept of model welfare without evidence.

    These rules on chatbot “welfare” follow co-founder Chris Olah’s warnings to religious scholars that he believed the…

    Used byThe Daily Caller (Far Right)

Show 7 more phrasesShow fewer

  • Loaded modifier

    ClankersMore neutral: AI models

    A slang term for robots that mocks the rule instead of describing it.

    Anthropic Launches New ‘Don’t Be Mean To The Clankers’ Policy

    Used byThe Daily Caller (Far Right)

  • Loaded modifier

    tech overlordsMore neutral: AI companies

    The label casts the companies as domineering without evidence in the passage.

    Our tech overlords have moved on from the systematic destruction of humanity to a new crisis: people…

    Used byThe Daily Caller (Far Right)

  • Sensational language

    systematic destruction of humanityMore neutral: risks that advanced AI could pose to humanity

    The phrase presents a speculative risk as an established fact in dramatic terms.

    Our tech overlords have moved on from the systematic destruction of humanity to a new crisis: people being mean to their chatbots.

    Used byThe Daily Caller (Far Right)

  • Sensational language

    RetardMore neutral: a derogatory insult

    The slur is used for shock in the headline and adds no information about the policy.

    DUKE: I Reserve The Right To Call My Chatbot A ‘Retard’

    Used byThe Daily Caller (Far Right)

  • Vague attribution

    critics saidMore neutral: some commentators said

    Attributes the concern to unnamed people, so readers cannot judge who holds it.

    But critics said the new rule risks treating an AI chatbot too much like a person, while supporters…

    Used byBusiness Insider (Left)

  • Vague attribution

    people are sayingMore neutral: name who

    Unfalsifiable attribution.

    Here's what smart people are saying about its move to ban abusive behavior.

    Used byBusiness Insider (Left)

  • Loaded verb

    BullyMore neutral: Abuse

    The headline recasts users’ rude messages as bullying, a judgement the policy text does not make.

    AI Giant Bans Users Who Bully Its Chatbot

    Used byThe Daily Beast (Far Left)

See the full language guide

See for yourself

Side by side

What each side leads with and what it leaves out. Open a side to read its headlines.

From the left

3 outlets

Leads with

  • users risk being banned for abuse of the chatbot
  • the rule is narrow and applies only in extreme cases
  • unclear how Anthropic will enforce the rule
  • Anthropic’s uncertainty about AI moral status

Leaves out

  • the election, weapons and surveillance restrictions in the same update

Reported by other sides, not found in this side’s coverage.

See the 3 headlines from the leftHide the headlines

From the center

1 outlet

Leads with

  • the ban on sustained and needless abusive behavior
  • the company’s own statement
See the headline from the centerHide the headlines

From the right

5 outlets

Leads with

  • mockery of being nice to chatbots
  • whether AI can feel or be harmed
  • AI as a risk to humanity
  • election, weapons and surveillance restrictions

Leaves out

  • the November 12 effective date

Reported by other sides, not found in this side’s coverage.

See the 5 headlines from the rightHide the headlines

Articles you open from here build My Mix, a private picture of how your reading spreads across the spectrum, kept on this device. What is My Mix?

The left is Far Left and Left together. The right is Right and Far Right together. Every outlet keeps its own five-point rating.

Who reported it when

How the coverage built up

  • Left

    13 hr 29 min later

    Its first report, Oct 9, 7:58 AM EDT

  • Center

    14 hr 5 min later

    Its first report, Oct 9, 8:34 AM EDT

  • Right

    First to report

    New York Post, Oct 8, 6:29 PM EDT

See the order the outlets reported inHide the order
  1. New York PostRightOct 8, 6:29 PM EDT
  2. The Daily CallerFar RightOct 8, 7:05 PM EDT
  3. Business InsiderLeftOct 9, 7:58 AM EDT
  4. BBC NewsCenterOct 9, 8:34 AM EDT
  5. The Daily BeastFar LeftOct 9, 10:45 AM EDT
  6. CBS NewsLeftOct 9, 10:59 AM EDT
  7. Breitbart NewsFar RightOct 9, 12:14 PM EDT
  8. The Washington TimesRightOct 9, 1:48 PM EDT
  9. ReasonRightOct 9, 4:05 PM EDT

Times are the ones the outlets published in their own feeds, in Eastern time. Opinion is left out.

Read next

Help someone else find the truth beneath the surface

Most people see one side of a story. Share this one, so they can see how every side is telling it and decide for themselves.

Email it

Seen a mistake? Tell us what is wrong with this page