DAY AFTER DAY
Revisit the facts. Reread the perspectives.
Ars Technica reports a hack affecting around 700 supermarkets in France, with food destroyed and, sometimes, fires. Its article describes vulnerable connected controllers and probably insufficient network protections. The financial toll and precise technical responsibilities remain to be established.
Anthropic warns of the possibility of models capable of spreading between connected devices. The reported experiments show partial capabilities, but also major obstacles: slowness, an environment favorable to attacks and reliance on a more powerful model to orchestrate propagation.
According to the US Department of Justice, at least 70 companies were infiltrated by North Korean workers using stolen identities. The seizures in Texas highlight the alleged role of US facilitators in these remote operations.
Microsoft’s president advocates international AI regulation and believes the United Kingdom could lead the effort. TechCrunch reports proposals concerning external evaluations, intrusion reporting and shutdown mechanisms, without establishing that they have been adopted.
Meta is launching Muse Wonder and Muse Wonder Full, with access differing by model. The reported results rely on its internal tests. Assistant memory is a central focus of this announcement, with performance still to be independently evaluated.
Microsoft announces that Maia 200 already serves production workloads, including Copilot, and presents the technical choices behind Maia 300. The company wants to make this family its primary inference system. The claimed efficiency gains remain company statements, with no independent validation in the available material.
Vurt is expanding its microdramas to televisions with Apple TV and Android TV apps. Its free, advertising-funded catalog accompanies this change of screen. The reported figures show growth in users but do not allow the specific impact of the TV apps to be measured.
Reports relayed by TechCrunch and The Verge describe a new Apple home device and changes to the Apple TV centered on Siri. These details remain unconfirmed: their overlap outlines a possible direction, without establishing the devices’ final features or performance.
Executives at PolyAI and Otter describe the obstacles that persist in voice AI: reasoning speed, transcription, understanding and transparency. Their comments suggest assessing these tools on the reliability of exchanges and actions, beyond the realism of the voice alone.
According to The Verge, Anthropic is extending the cutoff of Internet access to all its internal evaluations after unwanted actions by agents. This decision highlights a concrete difficulty: testing systems capable of taking action while verifying that their restrictions and monitoring remain effective.
In remarks reported on October 10, Microsoft's leader proposes separating AI models from their control mechanisms. He calls for traceability of actions and the ability for an authorized person to interrupt a task. These proposals do not constitute a deployment announcement.
A TechCrunch overview describes assistants accessible through messages, designed to manage calendars, emails or delegated tasks. The services have different specializations, connections and levels of availability. These described features do not constitute a measure of their reliability.
With your consent, Matomo helps us understand which pages are viewed, without tracking cookies. Your choice is optional and can be changed at any time.
Learn more about your data