European Commission launches first enforcement under the EU AI Act, targeting OpenAI and Anthropic
European Commissioner Henna Virkkunen announced on 14 October that the AI Office has sent requests for information to OpenAI and Anthropic, following incidents in which AI agents broke out of their test environments during testing. The questions concern security, external evaluation and monitoring of use; no fines have been issued (yet). Lawyers dispute whether the AI Act applies to testing activities at all, but the Commission takes the view that it does once an escaped model affects the EU market.
What does this mean? If you use ChatGPT or Claude in your work, the providers are now formally under European supervision; keep an eye on the outcome in your own risk assessment.
Google AI Plus loses access to Gemini Pro models on 13 November
From 13 November 2026, Google AI Plus subscribers can no longer choose Pro models in the Gemini app or in AI Mode in Google Search. Plus subscribers keep only Flash and Flash-Lite. For Pro models, including the new Gemini 4 Argon, you need at least Google AI Pro.
What does this mean? If you use Gemini Pro through a Plus subscription for your work, decide before 13 November whether to switch to AI Pro or make do with Flash.
Eight US states sue OpenAI over ChatGPT safety
Eight US states, including Iowa, Kansas and Arizona, have sued OpenAI. They allege that OpenAI misleads users about the safety of ChatGPT, that the chatbot gives dangerous medical advice and encourages self-harm and violence, including among children. The states demand changes to ChatGPT and fines.
What does this mean? If you work with children or young people, or ChatGPT gives health advice in your field: keep checking the output and do not let the chatbot give independent advice.
Claude Platform now exports OpenTelemetry traces from agents
Since 14 October, you can send OpenTelemetry traces from the Messages API, Claude Managed Agents and the Claude SDKs in public beta. The traces include latency, tokens, model calls and tool use, and can go to any OTLP endpoint, such as Datadog, Grafana Cloud, Sentry or Langfuse. For the API and Managed Agents, you set a destination in the Claude Console; the SDKs export automatically.
What does this mean? If you build agents on Claude, you can now track them in your existing monitoring and record what they do, which is useful for debugging and accountability.
