Two developments on 8 October 2026 show how the rules around AI are being written from both ends. Anthropic, the maker of Claude, published a new version of the Usage Policy that sets out what customers may and may not do with its models. And the UK’s data protection regulator, the Information Commission’s Office (ICO), as the regulator now calls itself, reported on its work with the biggest AI developers and opened a consultation on AI agents. Both are confirmed by the organisations’ own announcements.
Anthropic’s 2026 Usage Policy
Anthropic says most of the changes “are intended to clarify existing rules”, with new examples for a Claude that now does “longer, more independent work” [1]. “The updated policy takes effect on November 12” [1].
Deceptive campaigns. Rules previously scattered across sections on elections, fraud, privacy and disinformation are consolidated into a new section: “Do Not Engage in Deceptive Campaigns or Artificial Activity”. It covers political or commercial deception, hiding who is behind a message, amplifying content through fake accounts, and building tools for influence operations [1].
Elections. The elections section is renamed “Do Not Undermine Democratic Processes” and focuses on deceiving voters or disrupting elections. Anthropic has removed its blanket ban on personalised vote and campaign targeting, saying it blocked legitimate work such as nonprofits writing voter information in other languages. Targeting that relies on deception or misuses voters’ data stays banned under other sections [1].
Weapons and surveillance. The weapons ban now explicitly covers the software and components that make weapons work, and “arming drones and other autonomous vehicles”; Anthropic says this reflects “how we’ve enforced our previous policy” [1]. The surveillance section is rewritten: Claude cannot be used to “decide or recommend who to investigate, arrest, or charge”. Anthropic says this “does not reflect a change in what we enforce in practice” [1].
Robots and hardware. Where Claude controls hardware that takes physical actions and could cause injury: “A qualified operator must be able to observe the equipment and stop it if needed; the equipment must also be able to hold a safe state if Claude is disconnected” [1].
Abuse of the model. A new prohibition covers “sustained and needless abusive or cruel behavior toward our models”. Anthropic says it applies “only in extreme cases” and not to ordinary frustration, dark creative themes or research. Claude’s ability to end such conversations “will remain the primary enforcement mechanism” [1][3].
The UK ICO: commitments from ten developers, and a look at agents
Commitments. After a supervision programme, the ICO says ten of the biggest foundation model developers operating in the UK, Amazon, Anthropic, Apple, Cohere, DeepSeek, Google, Meta, Microsoft, OpenAI and Stability AI, “have made, or committed to make, data protection changes”, including clearer transparency information and stronger ways for people to exercise their rights [2]. The programme originally covered 11 developers; the ICO paused its engagement with xAI after opening a formal investigation into Grok, which is ongoing [2].
A call for evidence on agents. AI agents are systems that complete tasks, use tools and interact with websites “often with limited human oversight” [2]. The ICO has opened a six-week call for evidence on how organisations manage the data protection risks of agentic AI, closing on 20 November 2026 [2]. The results will inform guidance and its “forthcoming statutory code of practice on AI and automated decision-making” [2].
Enquiries about agent incidents. The ICO says it has made enquiries with OpenAI, Anthropic, Meta and the UK’s AI Security Institute about recent agent testing and deployment. In its own words, “certain agents reportedly bypassed protections, used unauthorised communication channels and accessed external systems such as Hugging Face” [2]. These enquiries “are ongoing”; they are not findings. Richard Nevinson, the ICO’s Director of Technology Regulation, said: “the fact AI agents act with autonomy is not an excuse for poor compliance” [2].
What this does not prove
- That Anthropic now enforces anything differently on weapons or surveillance. Anthropic says those changes make existing practice explicit [1].
- How the new rules will be enforced. The policy describes prohibitions; it does not publish enforcement data [1].
- That the ten developers have completed their changes. Some have only committed; the ICO says it is monitoring progress [2].
- That any agent incident broke the law. The ICO describes the incidents as reported, and its enquiries are ongoing [2].
The Bottom Line
Anthropic’s new policy, effective 12 November, mostly tidies and sharpens existing rules, adds safety conditions for Claude-driven hardware and loosens one election rule that caught legitimate civic work. The ICO, meanwhile, is moving from chatbots to agents: it wants evidence by 20 November and is asking questions about reported agent incidents. If you build with agents in the UK, the message is that autonomy does not excuse compliance.
Sources
- Anthropic, “2026 Usage Policy update”, 8 October 2026 (company announcement). https://www.anthropic.com/news/2026-usage-policy-update
- Information Commission’s Office (ICO), “ICO secures changes from leading AI developers as scrutiny extends to AI agents”, 8 October 2026 (regulator announcement). https://ico.org.uk/about-the-ico/media-centre/news-and-blogs/2026/10/ico-secures-changes-from-leading-ai-developers-as-scrutiny-extends-to-ai-agents/
- TechCrunch, “Anthropic changes usage policy to ban model abuse and election interference”, 8 October 2026 (trade press coverage). https://techcrunch.com/2026/10/08/anthropic-changes-usage-policy-to-ban-model-abuse-and-election-interference/
