Global AI, Malaysia Angle / AI news for Malaysia
Anthropic's new Claude usage policy takes effect 12 November with hardware and deception rules
Anthropic says most changes clarify existing rules, but firms using Claude in health, finance, hardware or cross-border ownership structures should read the new text before 12 November. The company names no Malaysia-specific change.
In brief
- Anthropic published its 2026 Usage Policy update on 8 October 2026. The new version takes effect on 12 November.[1]
- Changes include a new section on deceptive campaigns, a narrower elections rule, explicit weapons and surveillance wording, and controls for Claude connected to hardware that acts physically.[1]
- Anthropic says the high-risk requirements, a qualified human reviewer and disclosure that AI was used, are unchanged but now more clearly listed.[1]
What Anthropic changed
Anthropic published an updated Usage Policy on 8 October 2026 and says it takes effect on 12 November. The company says it updates the policy each year, and that most changes in this version clarify existing rules as Claude takes on longer, more independent work.[1]
This is a rules change by a supplier, not a law. It applies to anyone using Claude, including Malaysian businesses, through the supplier's terms.[1]
Deception, elections and surveillance
A new section, Do Not Engage in Deceptive Campaigns or Artificial Activity, gathers rules that were scattered across several sections. It covers hiding who is behind a message and amplifying content through fake accounts or posts, whether the purpose is political or commercial.[1]
The elections section is renamed Do Not Undermine Democratic Processes and now targets deceiving voters or disrupting elections. Anthropic removed its blanket ban on personalised vote and campaign targeting, because it also caught civic work such as voter information in other languages. Deceptive targeting stays banned elsewhere.[1]
The surveillance wording is sharper. Tracking people without consent is prohibited, in real time or from stored data, and Claude cannot decide or recommend who to investigate, arrest or charge. Consented tracking such as fraud monitoring, content moderation, journalism and legal research remain permitted.[1]
Weapons, hardware and high-risk use
The weapons section now says prohibitions include the guidance and control software and components that make weapons work, and actions such as arming drones and other autonomous vehicles. Anthropic says this reflects how it already enforced the policy.[1]
For high-risk uses affecting health, legal rights, finances, livelihood or access to essential services, a qualified human must be able to review and change Claude's recommendations, and affected people must be told AI was used. Anthropic says these requirements are unchanged, but the section now lists which recommendations are covered.[1]
A new control applies when Claude is connected to hardware that acts autonomously and could cause injury. A qualified operator must be able to observe the equipment and stop it, and the equipment must hold a safe state if Claude is disconnected.[1]
Abusive behaviour and supported regions
Anthropic added a ban on sustained and needless cruelty toward its models, aimed at extreme cases with no discernible purpose. It says frustration, pushback, dark creative themes and testing are not covered. Claude's ability to end such conversations remains the main enforcement tool.[1]
The Supported Regions Policy is clarified, not changed. Claude may not be used by people physically in an unsupported region, by entities incorporated or headquartered there, or by entities majority-owned or controlled from there. Malaysia appears on Anthropic's supported countries page, which we read on 11 October.[1][2]
Why Malaysia should care
Malaysia is on Anthropic's supported countries list, so Malaysian firms are not blocked by location. The ownership rule is different: entities majority-owned or controlled from an unsupported region fall outside the policy, wherever they operate.
Firms using Claude for health, finance or hiring decisions
A qualified person must be able to review recommendations, and affected people must be told AI was used.[1]
Practical move: Document who reviews outputs and how customers are told.
Companies with foreign shareholders or parents
Majority ownership or control from an unsupported region puts an entity outside the policy.[1][2]
Practical move: Check your ownership chain against the Supported Regions page.
Marketing and social media teams
Fake accounts and hidden-origin amplification are now grouped in one explicit section, for commercial use too.[1]
Practical move: Review any automated posting or review-writing workflows.
What Malaysians can do now
- Compliance and operations leads: list every workflow where Claude's output affects health, money, jobs or services, and name the human reviewer.
- Company secretaries and finance teams: compare your shareholding structure with Anthropic's Supported Regions page before 12 November.
- Teams connecting Claude to machines or drones: confirm an operator can observe and stop the equipment, and that it holds a safe state if disconnected.
What we still do not know
What we still do not know
- How Anthropic will enforce the policy beyond ending abusive conversations, and what notice it gives before closing an account.
- Which regions the Supported Regions Policy treats as unsupported in each ownership case, since the announcement does not list them.
- Whether Malaysian regulators will issue guidance on supplier AI policies, since none is cited in the announcement.
Sources
- 1.2026 Usage Policy update Anthropic, 8 October 2026
- 2.Supported countries and regions Anthropic
- 3.File:2026 Dewan Rakyat interior 1.jpg Wikimedia Commons, 12 September 2026
- 4.File:Quadcopter camera drone in flight.jpg Wikimedia Commons, 14 April 2018
- 5.File:Kuala Lumpur Malaysia Skyline-03.jpg Wikimedia Commons, 8 August 2012
More AI News
View all
