Lumma Stealer, a notorious infostealer that infected nearly 395,000 Windows computers before being disrupted last year, has resurfaced with renewed vigor, according to researchers. Simultaneously, advancements in artificial intelligence continue to emerge, with Chinese AI startup z.ai's GLM-5 achieving a record-low hallucination rate and MIT researchers developing a new fine-tuning method for large language models. In other news, the open-source AI assistant OpenClaw has been updated with a more secure version, NanoClaw. Finally, a call for a targeted economic strike against companies enabling certain government actions has been issued.
Lumma Stealer, also known as Lumma Stealer, first appeared in Russian-speaking cybercrime forums in 2022. Its malware-as-a-service model provided a sprawling infrastructure of domains for hosting lure sites offering free cracked software, games, and pirated movies, according to Ars Technica. After law enforcement authorities hobbled its infrastructure last May, the malware is now back in circulation, employing hard-to-detect attacks to pilfer credentials and sensitive files, researchers reported Wednesday.
In the realm of AI, z.ai's GLM-5, the latest in its GLM series, retains an open-source MIT License. VentureBeat reported that it achieves a record-low hallucination rate on the independent Artificial Analysis Intelligence Index v4.0, scoring -1 on the AA-Omniscience Index. This represents a 35-point improvement over its predecessor, positioning GLM-5 ahead of competitors like Google, OpenAI, and Anthropic in knowledge reliability.
Meanwhile, researchers at MIT, the Improbable AI Lab, and ETH Zurich have developed a new technique to allow large language models to learn new skills and knowledge without forgetting their past capabilities. Their technique, called self-distillation fine-tuning (SDFT), allows models to learn directly from demonstrations and their own experiments by leveraging the inherent in-context learning abilities of modern LLMs, according to VentureBeat. Experiments show that SDFT consistently outperforms traditional supervised fine-tuning.
The rapid adoption of Austrian developer Peter Steinberger's open-source AI assistant OpenClaw has led to the release of NanoClaw, a lighter, more secure version. OpenClaw, which offers a powerful means of autonomously completing work and performing tasks across a user's entire computer, phone, or even business with natural language prompts, has captured the market with over 50 modules and broad integrations since its release in November 2025. NanoClaw debuted under an open-source MIT license, according to VentureBeat.
Finally, a call for a targeted, month-long national economic strike has been issued on Hacker News. The strike aims to target tech companies and firms enabling certain government actions, with the goal of inflicting maximum damage with minimal impact on consumers. The strategy is based on the belief that the market is the primary driver of change.
Discussion
AI Experts & Community
Be the first to comment