Anthropic AI Updates: October 15, 2026

1. Claude Platform Adds OpenTelemetry Tracing in Public Beta

Anthropic. Anthropic launched OpenTelemetry tracing for Claude Managed Agents, the Messages API, and the Claude SDKs in public beta. Traces cover agent turns, model calls, tool use, and MCP server calls, each with latency, token counts, and IDs, and follow the OpenTelemetry GenAI semantic conventions so Claude data can share dashboards with other providers. Managed Agents and Messages API users add an OTLP export destination in the Claude Console, while the SDKs export traces automatically; named integrations include Grafana Cloud, Honeycomb, Sentry, Splunk, Datadog, Langfuse, LangSmith, Braintrust, Arize AI, and Databricks Unity Gateway. For teams running agents in production, this removes the need for custom instrumentation to see where latency and tokens go across tool and MCP calls. Source

2. Claude Code 2.1.300 Aligns Telemetry With GenAI Conventions and Expands Mod Hooks

Anthropic. Claude Code v2.1.300 adds OpenTelemetry GenAI semantic-convention attributes to trace spans, such as gen_ai.provider.name, gen_ai.conversation.id, gen_ai.tool.name, and error.type, and splits cache writes in the api_request event into cache_creation_5m_tokens and cache_creation_1h_tokens. The release brings /advisor to Claude Platform on AWS, off until a model is picked via /advisor, advisorModel, or --advisor. Mod authors get a prompt.change hook, the ability to rewrite a workflow agent’s prompt and model in the agent.spawn hook, $.ui.view({ agentId }) for showing an agent transcript, virtualText prompt decorations, and a dir prop for right-to-left text; the VS Code extension adds a claudeCode.attachSelection setting. Source

3. Anthropic Details Election Safeguards Ahead of the 2026 US Midterms

Anthropic. Anthropic reported that in a sample of 250,000 US Claude.ai conversations from September 21 to 27, about 0.51% concerned elections or voting and about 0.002% violated its election integrity policies. Later in October, Claude will pull official, state-specific voting information from Democracy Works into responses and will show official vote counts on and after election night. Claude Opus 5.5, Sonnet 5.5, and Haiku 5.5 scored 99% on Anthropic’s political evenhandedness evaluation, though external reviewers found uneven hedging and unsolicited counterarguments that Anthropic says it will address. Red-teaming with ISD and the McCain Institute showed Claude could miss influence campaigns split into many harmless-looking subtasks, so Anthropic added classifiers for multi-step misuse such as fake news sites and social media personas. Source

4. Anthropic Opens Singapore Office and Signs MOU With MDDI

Anthropic. Anthropic opened an office in Singapore, its fifth in Asia-Pacific, and signed a memorandum of understanding with the Ministry of Digital Development and Information covering AI safety and model evaluation, research on AI adoption, and responsible deployment in the public and private sectors. Planned work includes technical exchanges with Singapore’s AI assurance institutions on testing standards and incident response, SME adoption support, and sharing Anthropic Economic Index data with the government. Anthropic says Singapore ranks second of 121 countries in per-capita Claude.ai usage and that more than 1,500 GovTech Singapore engineers and technical staff use Claude Code to modernize government systems. Source