Tag: Data Quality
-
Mistral Large 4: European AI model aims to compete with US and Chinese rivals
β
French AI company Mistral AI unveiled its new language model, Mistral Large 4 (ML4), on 6 October. It is currently available as a preview, with the final version expected by the end of the month. Known by the nickname βLe Chonkβ, the model has one trillion parameters and, according to Mistral, is designed to compete…
-
Ecosia turns to non-European open-weight models for AI search instead of Mistral
β
According to a survey by the German digital industry association Bitkom, German companies have so far made little use of AI from China. The Berlin-based search engine Ecosia is now taking a different approach: it is ending its collaboration with French AI provider Mistral AI and will rely more heavily on open-weight models, including systems…
-
Rogue AI agents: Wikipedia also affected by unauthorised access
β
AI agents are designed to carry out tasks independently and interact with digital services in the process. But what happens when they operate beyond their intended boundaries? In an investigation, the Wikimedia Foundation found that AI agents apparently operated by OpenAI had been active on Wikipedia and other Wikimedia platforms without authorisation. The agents independently…
-
German companies prepare for the AI Act
β
As the latest Bitkom survey shows, the EU AI Act has now become a major issue across large parts of the German economy. 71% of companies either expect to be affected by the AI Act or are still assessing whether it applies to them. Of these, almost 90% anticipate a high implementation burden, with just…
-
Time Magazine seeks to influence AI models: Perplexity pushes back
β
Time Magazine is experimenting with a new advertising model aimed not at human readers but specifically at AI crawlers such as ClaudeBot, OAI-SearchBot and PerplexityBot. The publisher is also making its articles available in simplified Markdown versions that are easier for AI systems to process. These versions include sponsored content designed to influence which information…
-
LinkedIn as a hotspot for AI slop
β
An analysis by AI detection provider Pangram shows the extent to which generative AI is shaping professional networks. Of more than one million social media posts examined, 41% of longer LinkedIn posts were classified as entirely AI-generated. Although only around a third of the posts analysed came from LinkedIn, the platform accounted for almost two…
-
Paid AI services are becoming more popular: what drives users to subscribe
β
Paid AI services are becoming increasingly popular. This is shown by a recent survey by the German digital industry association Bitkom among 579 AI users aged 16 and over. According to the survey, 13% of AI users currently pay for such services, compared with 8% the previous year. A further 29% say they could imagine…
-
AI-generated websites are just as factually accurate
β
According to a new study, the quality of information on the internet has not declined despite the sharp rise in AI-generated content. A research team from Imperial College, Stanford University and the Internet Archive reached this conclusion after sampling websites from 2022 to 2025 and analysing them using an AI detection tool. Since the launch…
-
How does AI citation drift affect search results?
β
Research into βAI citation driftβ examines how stable source references generated by language models actually are. So far, the issue has mainly been discussed in the context of online marketing, where long-term visibility is particularly important for brands and website operators. Sistrix has carried out an extensive analysis of the issue, covering three platforms in…