LecturaInvestigaciónRadarMarco de inversión
Iniciar sesión / Registrarse
TodoTecnologíaProductosInversiónBolsaOtros

Lectura

2026-09-013 publicaciones

Actualizar

European Commission Places ChatGPT, Reddit and Roblox Under Very Large Service Oversight

Verificado 2026-09-01 11:24 GMT+8

On August 31, under the Digital Services Act, the European Commission designated ChatGPT as a “very large online search engine” and Reddit and Roblox as “very large online platforms.” All three services have at least 45 million average monthly users in the EU. Within four months of designation, they must submit systemic risk assessments, undergo external audits, and expand researcher data access and transparency disclosures; the Commission set the deadline at the end of November.

Regulation of ChatGPT had previously focused largely on model training, copyright, and the AI Act. This designation adds ongoing obligations for large-scale search and content distribution, bringing compliance costs tied to user scale into auditing, data governance, and risk reporting. The designation itself is not a penalty. The risk assessments due before the end of November will provide the first actionable guidance.

Source check European Commission designation announcement ↗ · Reuters report ↗

Fuentes:欧盟委员会指定公告

Investigación

Anthropic Extends Claude's Cross-Conversation Memory to Personal Subscribers

Verificado 2026-09-01 11:24 GMT+8

Anthropic extended the memory feature for Chat and cloud-based Cowork to Free, Pro, and Max plans on August 25. Users can view and edit memories by topic. Sensitive topics require explicit opt-in; personal plans enable memory by default, while Team and Enterprise plans are off by default. The product can therefore retain preferences and work context, but will not automatically access all past conversations. Enterprise adoption remains subject to admin policies, sensitive data scope, and deletion mechanisms.

Source verification Claude Release Notes ↗ · Memory Usage Guide ↗

Fuentes:Claude发布记录

Investigación

Astronomy alert field extraction hits 99.8% agreement with audit

Verificado 2026-09-01 11:24 GMT+8

Field extraction from astronomy alerts can now reach about 99.80% agreement with human audit: the Astro-COLIBRI team processed 1,775 astronomy alerts, and 25,827 of 25,880 explicit field judgments matched an internal human audit, with errors concentrated in observation times and facility attribution.

Previously this kind of alert field curation relied on manual work, which was costly and hard to scale. The method chains a constrained schema, model extraction and deterministic rule checks together.

The agreement rate is about 99.80%, measured by the team's own internal human audit (first-party self-test). Boundary: it covers only one scientific domain, has not been peer-reviewed or externally reproduced, and cannot be extrapolated to the reliability of general enterprise agents; the preprint was released on August 24 (arXiv:2608.23270).

Fuentes:论文摘要与全文入口

Investigación

Estás al día en esta vista

Iniciar sesión / Registrarse
LecturaInvestigaciónRadarMarco de inversión