Our method: distinguishing facts, citing sources and calmly explaining what can really change for creators, independent and digital professionals.
IA
Meta tests human operators for certain calls from Muse
Meta confirmed to Reuters that an internal Muse test involved human providers to support certain telephone calls requested from the agent. The function is not publicly deployed in this form: Meta indicates that it wants to collect returns, strengthen security and privacy protections, and then launch a possible version only with appropriate mentions. The test concerns the United States and recalls that a service presented as automated can still be based on human intervention.
What it can change: Before entrusting customer calls, reservations or exchanges to an AI agent, check who is actually performing the task, what data are transmitted and whether human intervention is clearly announced.
Read the source: Reuters, with official declaration of Meta →
Set-up
OpenAI and Anthropic offer a conditional Australian framework for AI training
In contributions filed with an Australian parliamentary inquiry, OpenAI and Anthropic call for a more permissive framework to train models from protected or publicly accessible content. Anthropic evokes a close authorisation with favourable conditions for creators; OpenAI defends a balance allowing learning about public information and collaborations with the beneficiaries. Australia has not adopted these proposals and has so far rejected a general copyright exemption; The Commission's report is expected in November.
What it can change: No new duty is applicable today. However, creators must monitor future consent, licensing, remuneration and withdrawal mechanisms, as this debate can inspire other countries.
Read the source: Reuters, based on contributions to the Australian Parliament →