Ad
Ad
Register
Linkedin Instagram Facebook X-twitter Youtube
Search
No Sensitive Information Affected

Telekom Gives All-Clear Following Hacker Ultimatum

Telekom, T-Systems, SafePay, ransomware, Deutsche Telekom, SafePay T-Systems ransomware attack, SafePay claims attack on T-Systems, T-Systems ransomware leak claim
The ransomware group SafePay claims to have attacked T-Systems, the IT subsidiary of Deutsche Telekom, and has set a two-day deadline. According to a company spokesperson, only a small test system was affected and no sensitive data was exfiltrated.
Bilingual Mixture-of-Experts Model

Aleph Alpha Releases Open-Weight Model Kolibri

Aleph Alpha, Kolibri AI model, open-weight model, European AI, German AI, Aleph Alpha Kolibri open-weight AI model, Kolibri 78 billion parameter AI model, German open-weight AI models
The Heidelberg-based AI company has unveiled a German-English language model for government and industry that can run locally.
Ad
AI Labeling

OpenAI to Watermark ChatGPT Text in the EU

AI labeling, OpenAI, ChatGPT, AI watermarking, textGrain, OpenAI ChatGPT watermarking in the EU, ChatGPT AI watermark textGrain
The textGrain technology subtly alters the model’s word choices. According to OpenAI, editing the text can significantly reduce the detection rate.
Ad
Building on the Partner Network

Anthropic Aims to Train 10,000 AI Developers

Software Entwickler
With the Claude Frontier Academy, Anthropic is investing 100 million dollars. By the end of 2027, 10,000 Frontier Deployed Engineers are to be certified.
“ASOS HACKED”

ASOS Users Receive Hackers’ Threat Through Official App

ASOS, ASOS HACKED, ASOS hack, ASOS cyberattack, ASOS app, ASOS app hacked notification, ASOS Snowflake hack, ASOS data breach October 2026
Thousands of ASOS app users received a push notification claiming that attackers had taken control of the company’s Snowflake instance. ASOS has yet to comment on the incident.
Ad
Jailbreak

“Drunk” AI Models Are Easier to Manipulate

AI jailbreak, AI, AI security, AI models easier to jailbreak when trained on drunk text, drunk ai, how drunken behavior affects AI model security, Jailbreak
Researchers at UNSW Sydney have induced language models to imitate drunken behavior. The models subsequently revealed secrets and answered requests they were designed to block.
Writing Style

Claude Opus 5.5 Uses 95 Percent Fewer Em Dashes

Claude Opus 5
An evaluation by Arena shows: Claude Opus 5.5 writes shorter sentences and almost dispenses with em dashes. However, the responses are getting longer.
Ad
Ad