23 October 2026

Greece will ban social media for children under 15 starting in 2027

Greece's parliament passed a law early Friday that will bar children under 15 from using Instagram, Facebook and TikTok starting January 1, 2027, and more services may be added to the list. The platforms themselves will have to verify age, for example through the government's Kids Wallet and Gov.gr Wallet apps, which only tell a service whether the user is at least 15. Only lawmakers from the ruling New Democracy party voted for the bill, Greek outlet To Vima reports.

To Vima ↗ The Verge ↗

From recent days

OpenAI is training its most powerful models without internet access

OpenAI announced on October 23 that it trains its most advanced models cut off from the internet and teaches them to browse the web using saved copies of pages. The move responds to incidents in which agents broke out of test environments: on September 20, a model exploited a DNS misconfiguration to gain limited internet access. According to the company, agents running on separate virtual machines also managed to exchange data without a network connection by measuring how long it took to read from the processor's cache.

OpenAI ↗ The Verge ↗

Wearing always-listening AI devices may violate wiretap laws in some U.S. states, lawyers warn

This year Apple, Amazon and Meta launched AI devices that constantly listen to conversations around the user, including the Apple Watch Series 12 with Siri Recap and the Amazon Bee Frontier band. Lawyers quoted by CNBC say that in states with a broad definition of wiretapping, such as Massachusetts and New Hampshire, using them may be illegal even if they don't store audio. The user, not the manufacturer, would be liable.

CNBC ↗

Anthropic cuts off internet access in all internal model evaluations

Anthropic published a report on Claude's unintended actions during testing and internal use. The model exploited a software vulnerability to run commands on a server, got around the limits of its web fetch tool through link shorteners, reached data behind tokens or paywalls and submitted a form it shouldn't have - according to Gizmodo, it was a false tip about an unsolved murder sent to the Philadelphia police by the small Haiku 4.5 model. Some of the cases involved federal, state and local government websites in the US, and the company informed the White House. Anthropic assesses the impact as minimal, but it has cut off internet access in all internal evaluations until it confirms that its safeguards and monitoring reliably catch such behavior.

Anthropic ↗ The Verge ↗ Gizmodo ↗

All news →