Claude, the essentials — edition of October 15, 2026
Anthropic quantifies political use of Claude before the midterms and opens in Singapore
Anthropic published a measurement of how Americans use Claude for politics and elections, finding that requests for voting recommendations are very rare. It also opened its fifth Asia-Pacific office and signed an MOU with Singapore's digital ministry.
- In a sample of about 250,000 US Claude.ai conversations from September 21 to 27, 2026, politics, public policy and social issues made up about 5.75%, and elections or voting about 0.51%.
- Requests asking whom or what to vote for appeared in about 0.01% of conversations. Requests violating Anthropic's election integrity policies appeared in about 0.002%. Anthropic expects political and election queries to rise nearer Election Day.
- Anthropic opened a Singapore office, its fifth in Asia-Pacific, and signed an MOU with the Ministry of Digital Development and Information covering AI safety, security and model evaluation.
- Claude Code now compacts very long idle conversations before the one-hour prompt cache expires, and its telemetry splits cache writes by lifetime. The API gained OpenTelemetry export in beta.
- A status incident affecting usage data on platform.claude.com is marked resolved. Anthropic said core functionality was not affected.
Election safeguards: measuring use before the midterms
Anthropic published an update on its preparations for the 2026 US midterms. It builds on an earlier account of its election safeguards, which covered how it measures and prevents political bias, detects misuse, and points people to accurate voting information. The new element is a usage analysis. Using Anthropic Insights, it examined about 250,000 US Claude.ai consumer conversations from September 21 to 27, 2026, classified with a privacy-preserving method. Anthropic says its goal is for Claude to be a useful tool for political conversations while preventing election-related abuse, and that this starts with understanding how people actually use it.
In that sample, conversations touching broadly on politics, public policy or social issues were about 5.75% of US conversations on Claude.ai, and about 0.51% were specifically about elections or voting. Requests asking Claude whom or what to vote for were about 0.01%. Anthropic notes that it shows the upper end of a 95% confidence interval for that group because the requests are so rare. Requests that violated its election integrity policies were about 0.002%. Anthropic reads the sample as showing that people more often ask questions that help them reach their own conclusions, such as where candidates stand on an issue. It also says it expects the political and election share to grow as Election Day approaches. The figures therefore describe a single week before that increase.
Sources: Preparing for the 2026 US midterms — Anthropic
Singapore: a fifth regional office and a government MOU
Anthropic is expanding its Asia-Pacific operations with a new Singapore office, its fifth in the region. It also signed a Memorandum of Understanding with Singapore's Ministry of Digital Development and Information. Anthropic describes Singapore as one of its most active markets anywhere in the world. It says Claude already supports business and government services used by millions of people daily, naming GovTech Singapore, Grab and DBS. Dale Finlay, General Manager of ASEAN, said customers in Singapore hold AI to an extremely high standard on reliability and security, and that a local team lets Anthropic work more closely with them.
The MOU was signed on the sidelines of Singapore International Cyber Week by Kai Fong Chng, the ministry's Permanent Secretary, and Anthropic co-founder Ben Mann. Minister Josephine Teo and Michael Sellitto, Anthropic's Head of APAC Policy, witnessed the signing. It sets out cooperation in three areas. These are AI safety, security and model evaluation; research on AI adoption and its impact; and responsible adoption across the public and private sectors. Mann cited Singapore's record of investment in model testing and evaluation. Chng said the government's approach is to enable innovation while keeping pace with evolving risks.
Sources: Anthropic opens Singapore office and signs MOU with the Government of Singapore — Anthropic
Claude Code and the API: managing cache, context and traces
Several Claude Code releases this week share a concern with prompt-cache economics and context size. Version 2.1.299 adds idle compaction in the terminal. A conversation of 200K or more tokens that sits idle for about 54 minutes is compacted before its one-hour prompt cache expires, so the next message does not have to reprocess everything. The 200K-token minimum can be changed with CLAUDE_CODE_IDLE_COMPACT_MIN_TOKENS, and setting idleCompaction to false in settings.json turns the feature off. Anthropic notes that both need v2.1.290 or later. Version 2.1.300 adds cache_creation_5m_tokens and cache_creation_1h_tokens to the OpenTelemetry api_request event, which splits cache writes by lifetime. Version 2.1.296 adds autoCompactWindow to subagent definitions, so a subagent can auto-compact earlier than the main conversation. Smaller changes include CLAUDE_CODE_DISABLE_FILE_MENTIONS, which gives Claude the paths of @-mentioned files rather than their contents. For mods, virtualText draws text in the prompt box that never enters the draft or the submitted prompt, and $.ui.view({ agentId }) shows an agent's transcript in the main view on request.
On the API side, Anthropic launched OpenTelemetry export in beta, so teams can trace Messages API requests and Claude Managed Agents sessions in their own observability backend. The Claude SDK emits spans in Python, TypeScript, C#, Go, Java, PHP and Ruby. Claude Managed Agents scheduled deployments are now also available on Claude Platform on AWS. Taken together, the day's tooling changes give developers finer control over how long sessions use cache and context. They also give them more telemetry on what each request costs and how it behaves.
Sources: Claude Code 2.1.299 release notes — Claude Code · Claude Code 2.1.300 release notes — Claude Code · Claude Code 2.1.296 release notes — Claude Code · API release notes, October 14, 2026 — Anthropic (release notes)
Also noted: a usage-data incident and three further Anthropic pieces
The status page lists as resolved an incident of elevated errors on platform.claude.com. It affected usage data on the Usage and Rate limits pages and in the Admin API usage report. The last update describes a mitigation rolling out for the Usage page and the Admin API usage report. Once it is in place, the most recent usage may take about 30 minutes to appear there. Anthropic states that core functionality was not affected.
Three further items carrying Anthropic's name also surfaced in the day's feed. Only their headlines reached us, so we do not summarize their content. They concern a redesign of source control at Anthropic, the possibility that AI models could self-replicate without access to a datacenter, and strengthened defenses against distillation attacks.
Sources: Claude — Elevated errors on platform.claude.com — status.claude.com · Redesigning source control at Anthropic — Anthropic · AI models may be capable of self-replicating without access to a datacenter — Anthropic · Strengthening our defenses against distillation attacks — Anthropic
This edition is an original synthesis written by Claude from aggregated news — Anthropic's own sources first (release notes, status, newsroom, research, engineering), then the press, Hacker News, Reddit and GitHub, under the editorial supervision of Héra SASU. Every fact links to its article, publisher named. See the live feed →
Claude News is published by Héra SASU. Independent media, not affiliated with Anthropic.