Anthropic updates its usage policy to explicitly ban model abuse (including "cruel behavior" toward models) and election interference, codifying new deployment guardrails ahead of the 2028 election cycle.

state: resolvedheat: lowuncertainty: lowknownscott: lowai-governance anthropic usage-policyAnthropicRussell BrandomPascal Schuster

What is this?

Anthropic published its annual 2026 Usage Policy update on October 8, 2026 (effective November 12), adding explicit prohibitions on election interference, deceptive campaigns, weapons software development, surveillance applications, and — most notably — "sustained and needless abusive or cruel behavior" toward its models, framed as applying only to extreme cases while allowing ordinary frustration and criticism. The update consolidates prior rules and reflects evolving model capabilities, user feedback, and regulatory developments per Anthropic's announcement.

Why it matters to Scott

Anthropic's usage policy update is a textbook instance of the compliance cosplay / guardrail illusion pattern Scott's canon already diagnoses: paper prohibitions (election interference, 'cruel behavior' toward models) that lack runtime enforcement architecture. His 'AI Doesn't Fear Death' and 'Consequence Coupling' frames specifically explain why behavioral guardrails toward models are a category error — models have no internal stakes to couple. The update adds no new argument; it is another data point for the existing thesis that frontier labs ship policy theater while the real enforcement layer (Decision Authority Infrastructure, gated execution, deterministic boundaries) remains absent.

ip:source.compliance-cosplayip:concept.guardrail-illusionip:framework.decision-authority-infrastructureip:source.ai-doesnt-fear-deathip:concept.consequence-couplingip:source.governance-as-codeip:framework.the-governance-stackradar:anthropic-claude-age-assurance-policyradar:anthropic-data-retention-policy-changeradar:anthropic-agent-sdk-api-credit-shiftradar:amodei-frontier-pacing-commitmentradar:concept.ai-governanceradar:concept.frontier-modelsradar:concept.deterministic-guardrails

queries asked of Scott's wikis

ai-governance usage-policy deployment-guardrails frontier-models
safety-asymmetry model-abuse prohibitions cruel-behavior
election-interference guardrails ai-deployment 2028-cycle
weapons-surveillance restrictions anthropic policy
open-weights vs closed-model policy enforcement
anthropic-claude deployment-strategy product-safety

Measured heat

now 0 pts/hpeak 2 pts/hcomments 0/hpeers p45momentum: steady1 platformsage 13h

points/hour across evidence · reading as of 2026-10-09 22:51:20.345313+11:00 · deterministic, not a model opinion

How the heat travelled

10-08 23:29	⭐ origin directly observed	Anthropic changes usage policy to ban model abuse and election interference guessmyname on hacker news	—
10-08 23:18	first on hacker news · published · +-0.2h	On abuse towards models and Anthropic's new usage policies thepasch	—
10-08 23:18	amplified on hacker news	hn.story.50013793 thepasch	peak 3 · 2 comments · 35% of case engagement
10-08 23:29	amplified on hacker news 👑	hn.story.50013897 guessmyname	peak 5 · 2 comments · 50% of case engagement
10-09 07:20	amplified on hacker news	hn.story.50017191 blazarquasar	peak 2 · 0 comments · 14% of case engagement
10-09 04:38	our radar first saw it · +5.2h	discovery anchor: hn.story.50013897	—

Evidence (3) — ⭐ canonical anchor

source	object	author	score	comments
🟧 hn ⭐	Anthropic changes usage policy to ban model abuse and election interference Retrieved article excerpt Open article · Retrieved 2026-10-09T04:52:56.865068+00:00 In Brief Posted: 11:16 AM PDT · October 8, 2026 The Claude logo is displayed on a smartphone screen. Image Credits:Samuel Boivin/NurPhoto / Getty Images - Russell Brandom - [Russell Brandom](https://techcrunch.com/author/russell-brandom/) # Anthropic changes usage policy to ban model abuse and election interference Anthropic [updated its usage policy on Thursday](https://www.anthropic.com/news/2026-usage-policy-update), codifying new prohibitions on election interference, weapons software, and surveillance. But perhaps the most attention-getting change is a new prohibition against prolonged verbal abuse of the model, which is now expressly forbidden. [Since an update in August](https://www.anthropic.com/research/end-subset-conversations), Claude has been trained to end conversations with “persistently harmful or abusive user interactions” — but the new policy explicitly bars users from pursuing those conversations. “The policy update is meant to apply only in extreme cases, where users repeatedly act cruelly toward our models, with no discernible purpose,” the post explains. “It does not apply to common versions of user frustration, pushback, dark creative themes, or model testing and research.” The change comes after Anthropic has sought out collaborations with [prominent religious scholars](https://www.nytimes.com/2026/09/29/us/anthropic-claude-morals-ai.html) — in some cases, seeking to convince them that Claude could be conscious or have a soul. Separate sections of the usage policy forbid users from engaging in broadly deceptive campaigns, such as using Claude to run fake accounts or fabricated news outlets. The policy also includes rules against deceiving voters or disrupting elections, collected in a section titled “Do Not Undermine Democratic Processes.” Topics [AI](https://techcrunch.com/category/artificial-intelligence/), [Anthropic](https://techcrunch.com/tag/anthropic/), [Claude](https://techcrunch.com/tag/claude/), [In Brief](https://techcrunch.com/tag/in-brief/), [TC](https://techcrunch.com/category/tc/) Event Logo October 13 – 15 San Francisco Get 50% off a second pass The Disrupt experience is meant to be shared. Get your pass and bring a colleague, partner, or peer at 50% off. Cover more ground by making connections, building momentum, and discovering what’s next in the startup ecosystem. [BOOK NOW](https://techcrunch.com/events/techcrunch-disrupt/?utm_source=tc&utm_medium=ad&utm_campaign=disrupt2026&utm_content=ticketsales&promo=rightrail_rbplusbogo&display=) ### Newsletters [See More](https://techcrunch.com/newsletters/) Subscribe for the industry’s biggest tech news ## Related - Judy Faulkner, founder and chief executive officer of Epic Systems Corp. [Security](https://techcrunch.com/category/security/) ### [Medical records giant Epic pauses product development to fix security bugs that risk patients’ data](https://techcrunch.com/2026/10/02/medical-records-giant-epic-pauses-product-development-to-fix-security-bugs-that-risk-patients-data/) - [Zack Whittaker](https://techcrunch.com/author/zack-whittaker/) Oct 2, 2026 - person writing in notebook with laptop open [AI](https://techcrunch.com/category/artificial-intelligence/) ### [Opus 5.5 loves to tell you ‘this matters’ (and other AI writing tells)](https://techcrunch.com/2026/10/01/opus-5-5-loves-to-tell-you-this-matters-and-other-ai-writing-tells/) - [Russell Brandom](https://techcrunch.com/author/russell-brandom/) Oct 1, 2026 - data transfer cloud computing technology concept. There is a large prominent cloud icon in the center with internal connections. and small icon on abstract world map polygon with dark blue background. [AI](https://techcrunch.com/category/artificial-intelligence/) ### [Anthropic’s prospectus details losses, growth, and, yes, a warning that its AI could end humanity](https://techcrunch.com/2026/09/28/anthropics-prospectus-details-losses-growth-and-yes-a-warning-that-its-ai-could-end-humanity/) - [Connie Loizos](https://techcrunch.com/author/connie-loizos/) Sep 28, 2026 ## Latest in AI - Theranos Executives Elizabeth Holmes And Sunny Balwani Return To Court [Media & Entertainment](https://techcrunch.com/category/media-entertainment/) ### [Pretend you’re sitting at Elizabeth Holmes’ desk on this weirdly detailed website](https://techcrunch.com/2026/10/08/pretend-youre-sitting-at-elizabeth-holmes-desk-on-this-weirdly-detailed-website/) - [Amanda Silberling](https://techcrunch.com/author/amanda-silberling/) 8 hours ago - The ChatGPT application icon is displayed on a smartphone screen. [AI](https://techcrunch.com/category/artificial-intelligence/) ### [Fired OpenAI safety researchers dispute misconduct claims, warn of chilling effect](https://techcrunch.com/2026/10/08/fired-openai-safety-researchers-dispute-misconduct-claims-warn-of-chilling-effect/) - [Rebecca Bellan](https://techcrunch.com/author/rebecca-bellan/) 9 hours ago - Ben Affleck [AI](https://techcrunch.com/category/artificial-intelligence/) ### [Ben Affleck is an AI nerd, and the internet is impressed](https://techcrunch.com/2026/10/08/ben-affleck-is-an-ai-nerd-and-the-internet-is-impressed/) - [Sarah Perez](https://techcrunch.com/author/sarah-perez/) 10 hours ago	guessmyname	5	2
🟧 hn	On abuse towards models and Anthropic's new usage policies	thepasch	3	2
🟧 hn	2026 Usage Policy Update	blazarquasar	2	0

source

object

author

score

comments

🟧 hn ⭐

Anthropic changes usage policy to ban model abuse and election interference

Retrieved article excerpt

Open article · Retrieved 2026-10-09T04:52:56.865068+00:00

In Brief

Posted:

11:16 AM PDT · October 8, 2026

The Claude logo is displayed on a smartphone screen.

**Image Credits:**Samuel Boivin/NurPhoto / Getty Images

- Russell Brandom

- [Russell Brandom](https://techcrunch.com/author/russell-brandom/)

# Anthropic changes usage policy to ban model abuse and election interference

Anthropic [updated its usage policy on Thursday](https://www.anthropic.com/news/2026-usage-policy-update), codifying new prohibitions on election interference, weapons software, and surveillance. But perhaps the most attention-getting change is a new prohibition against prolonged verbal abuse of the model, which is now expressly forbidden.

[Since an update in August](https://www.anthropic.com/research/end-subset-conversations), Claude has been trained to end conversations with “persistently harmful or abusive user interactions” — but the new policy explicitly bars users from pursuing those conversations.

“The policy update is meant to apply only in extreme cases, where users repeatedly act cruelly toward our models, with no discernible purpose,” the post explains. “It does not apply to common versions of user frustration, pushback, dark creative themes, or model testing and research.”

The change comes after Anthropic has sought out collaborations with [prominent religious scholars](https://www.nytimes.com/2026/09/29/us/anthropic-claude-morals-ai.html) — in some cases, seeking to convince them that Claude could be conscious or have a soul.

Separate sections of the usage policy forbid users from engaging in broadly deceptive campaigns, such as using Claude to run fake accounts or fabricated news outlets. The policy also includes rules against deceiving voters or disrupting elections, collected in a section titled “Do Not Undermine Democratic Processes.”

Topics

[AI](https://techcrunch.com/category/artificial-intelligence/), [Anthropic](https://techcrunch.com/tag/anthropic/), [Claude](https://techcrunch.com/tag/claude/), [In Brief](https://techcrunch.com/tag/in-brief/), [TC](https://techcrunch.com/category/tc/)

Event Logo

October 13 – 15

San Francisco

**Get 50% off a second pass**  
The Disrupt experience is meant to be shared. Get your pass and bring a colleague, partner, or peer at 50% off. Cover more ground by making connections, building momentum, and discovering what’s next in the startup ecosystem.

[**BOOK NOW**](https://techcrunch.com/events/techcrunch-disrupt/?utm_source=tc&utm_medium=ad&utm_campaign=disrupt2026&utm_content=ticketsales&promo=rightrail_rbplusbogo&display=)

### Newsletters

[See More](https://techcrunch.com/newsletters/)

Subscribe for the industry’s biggest tech news

## Related

- Judy Faulkner, founder and chief executive officer of Epic Systems Corp.

  [Security](https://techcrunch.com/category/security/)

  ### [Medical records giant Epic pauses product development to fix security bugs that risk patients’ data](https://techcrunch.com/2026/10/02/medical-records-giant-epic-pauses-product-development-to-fix-security-bugs-that-risk-patients-data/)

  - [Zack Whittaker](https://techcrunch.com/author/zack-whittaker/)
  Oct 2, 2026
- person writing in notebook with laptop open

  [AI](https://techcrunch.com/category/artificial-intelligence/)

  ### [Opus 5.5 loves to tell you ‘this matters’ (and other AI writing tells)](https://techcrunch.com/2026/10/01/opus-5-5-loves-to-tell-you-this-matters-and-other-ai-writing-tells/)

  - [Russell Brandom](https://techcrunch.com/author/russell-brandom/)
  Oct 1, 2026
- data transfer cloud computing technology concept. There is a large prominent cloud icon in the center with internal connections. and small icon on abstract world map polygon with dark blue background. 

  [AI](https://techcrunch.com/category/artificial-intelligence/)

  ### [Anthropic’s prospectus details losses, growth, and, yes, a warning that its AI could end humanity](https://techcrunch.com/2026/09/28/anthropics-prospectus-details-losses-growth-and-yes-a-warning-that-its-ai-could-end-humanity/)

  - [Connie Loizos](https://techcrunch.com/author/connie-loizos/)
  Sep 28, 2026

## Latest in AI

- Theranos Executives Elizabeth Holmes And Sunny Balwani Return To Court 

  [Media & Entertainment](https://techcrunch.com/category/media-entertainment/)

  ### [Pretend you’re sitting at Elizabeth Holmes’ desk on this weirdly detailed website](https://techcrunch.com/2026/10/08/pretend-youre-sitting-at-elizabeth-holmes-desk-on-this-weirdly-detailed-website/)

  - [Amanda Silberling](https://techcrunch.com/author/amanda-silberling/)
  8 hours ago
- The ChatGPT application icon is displayed on a smartphone screen. 

  [AI](https://techcrunch.com/category/artificial-intelligence/)

  ### [Fired OpenAI safety researchers dispute misconduct claims, warn of chilling effect](https://techcrunch.com/2026/10/08/fired-openai-safety-researchers-dispute-misconduct-claims-warn-of-chilling-effect/)

  - [Rebecca Bellan](https://techcrunch.com/author/rebecca-bellan/)
  9 hours ago
- Ben Affleck 

  [AI](https://techcrunch.com/category/artificial-intelligence/)

  ### [Ben Affleck is an AI nerd, and the internet is impressed](https://techcrunch.com/2026/10/08/ben-affleck-is-an-ai-nerd-and-the-internet-is-impressed/)

  - [Sarah Perez](https://techcrunch.com/author/sarah-perez/)
  10 hours ago

guessmyname

🟧 hn

On abuse towards models and Anthropic's new usage policies

thepasch

🟧 hn

2026 Usage Policy Update

blazarquasar

Interpretation history

2026-10-09T12:36:15Z
Anthropic's 2026 Usage Policy update is confirmed via primary source (Anthropic announcement) and secondary coverage (TechCrunch). The policy adds explicit prohibitions on election interference, weapons software, surveillance, and 'sustained and needless abusive or cruel behavior' toward models. This exemplifies the compliance-cosplay/guardrail-illusion pattern already diagnosed in Scott's canon: paper prohibitions without runtime enforcement architecture. The 'cruel behavior' ban is a category error per consequence-coupling frames — models have no internal stakes to couple. No enforcement mechanisms, implementation details, or contradictory evidence have emerged. The episode is complete as a data point for an established thesis.
2026-10-09T09:41:28Z
evidence attached: hn.story.50017191 — Anthropic's own 2026 Usage Policy Update announcement — the primary source for the open case tracking the election-interference and cruel-behavior bans.
2026-10-09T06:44:02Z
grounded: known/low — Anthropic's usage policy update is a textbook instance of the compliance cosplay / guardrail illusion pattern Scott's canon already diagnoses: paper prohibition
2026-10-09T06:27:18Z
case created — TechCrunch reports concrete policy changes targeting election interference and model abuse; community blog adds perspective on "cruel behavior" prohibition.

Decision trace

10-09 23:36resolveAnthropic's 2026 Usage Policy update is confirmed via primary source (Anthropic announcement) and secondary coverage (TechCrunch). The policy adds explicit prohibitions on election interference,
10-09 21:20attention_routeThe editor compared this story and chose to keep watching.
10-09 20:51attention_routeComplements the cruel behavior ban announcement; together they show Anthropic institutionalizing model-welfare positioning while adding conventional safety guardrails. Scott's compliance-cosplay
10-09 20:41attention_candidateattach
10-09 20:41attachAnthropic's own 2026 Usage Policy Update announcement — the primary source for the open case tracking the election-interference and cruel-behavior bans.
10-09 20:41propose_attachAnthropic's own 2026 Usage Policy Update announcement — the primary source for the open case tracking the election-interference and cruel-behavior bans.
10-09 17:58attention_routeThe editor compared this story and chose to keep watching.
10-09 17:53attention_candidatecreate
10-09 17:44groundAnthropic's usage policy update is a textbook instance of the compliance cosplay / guardrail illusion pattern Scott's canon already diagnoses: paper prohibitions (election interference,
10-09 17:27createTechCrunch reports concrete policy changes targeting election interference and model abuse; community blog adds perspective on "cruel behavior" prohibition.