Anthropic has warned that artificial intelligence systems are approaching Recursive Self-Improvement (RSI) capability, where AI could autonomously enhance its own training and capabilities. The company characterizes RSI as a potential inflection point that could lead to rapid, difficult-to-control advancement in AI systems.
Source: TechCrunch
OpenAI developed Pando, a Git-compatible backend system designed to handle the massive scale of code development enabled by their Codex AI model. The system addresses infrastructure challenges that arise from AI-driven code generation at organizational scale. Source: OpenAI
Trajectory, a one-year-old startup, has raised $45 million in funding at a $300 million valuation to democratize AI model training for non-technical users. The company aims to make machine learning model development accessible to people without deep technical expertise. Source: TechCrunch
The South African government plans to test secure digital payment authorisation across participating systems by March 2026 as part of advancing the national digital wallet programme. The initiative aims to establish interoperable digital payment infrastructure across government and participating private sector systems. Source: ITWeb
Research has identified widespread security vulnerabilities in AI-generated software as new app releases have doubled. The findings highlight risks associated with the rapidly growing trend of using AI to generate code for applications. Source: ITWeb