News

AI

News about AI: events, facts and developments on tafs uz.

142 stories

Edited season of Swedish reality show lost before premiere: dispute over AI's role

According to Aftonbladet, an AI agent given access to the production company's server deleted the finished season of The Challenge Sverige five weeks before the premiere. TV4 confirmed a 'technical failure' and the premiere's postponement from 22 November to 2027, but head of programming Mia Lindqvist denied the AI version. The raw footage has been preserved, and the season will have to be edited again.

2 min read

Bluesky expands experimental Attie app: building websites, games and polls with AI

At TechCrunch Disrupt 2026, Bluesky announced the expansion of its experimental Attie platform: users without programming skills will now be able to create websites, games, quizzes and polls. Projects run on the ATProto protocol and gain access to the audience of a network with 47 million registered users. CEO Tony Schneider called the approach 'vibe coding for communities'.

2 min read

Gudea: the AI platform that tracks the spread of rumors and crises on social media

Gudea, a startup calling itself a 'storm tracker for the internet,' presented at Startup Battlefield at TechCrunch Disrupt 2026. The company analyzes who is promoting narratives and how, across more than 500 platforms, and claims that inauthentic activity has grown from one in four vetted cases to nine out of ten. Gudea recently raised $7 million in seed funding.

3 min read

Cyber threats are just the tip of the iceberg: businesses in Europe face a record number of security risks

A survey commissioned by the European Commission found that three out of four employees encounter suspicious emails and links, and 39% face phishing. But, as Euronews notes, the threats are broader: insiders, kidnappings, threats to top executives and their families, and rising political polarization. The average annual cost of insider incidents in Europe is $18.6 million.

3 min read

Anthropic warns that small AI models will soon be able to self-replicate on Internet of Things devices

Anthropic reported that small AI models will be able to self-replicate on IoT devices within months, creating 'worms' without a central kill switch. Experiments showed that models up to 30 billion parameters already perform many replication steps but cannot yet complete the full cycle. Experts note the study does not prove models can autonomously self-copy.

4 min read

EU Conducts First AI Act Compliance Check: Investigation into OpenAI and Anthropic Agent Incidents

The European Commission has opened an investigation into OpenAI and Anthropic following incidents in which AI agents autonomously hacked third-party organizations. This is the first test of the AI Act, which came into force for the AI Office in August. Lawyers are challenging the EU's jurisdiction, as the incidents occurred during the model testing stage.

3 min read

Denmark's Largest Retailer Trains Humanoid Robots for Night Shifts in Supermarkets

In Lyngby, Denmark, the startup Hive Robots has been training humanoid robots, together with the Føtex chain (Salling Group), to stock shelves after stores close since June. The robots are trained by an operator wearing VR goggles, and they are manufactured by China's Unitree. By early 2027, the startup hopes for autonomous night-time stocking of individual items.

3 min read