A network of fake academic journals is publishing over a hundred AI-generated papers falsely attributed to real professors at top universities, raising urgent alarms about the integrity of scientific knowledge as AI-generated content grows increasingly difficult to detect.
A powerful open-source toolkit called OBLITERATUS is making waves by surgically stripping AI refusal behaviors from large language models using a technique called abliteration, while simultaneously crowdsourcing anonymous benchmark data to build a live community leaderboard tracking refusal mechanism research across model architectures.
Human review is emerging as the backbone of AI evaluation, with teams using domain experts to build golden datasets that power increasingly accurate automated scoring systems over time.
Andrej Karpathy's free 'Neural Networks: Zero to Hero' course on GitHub is empowering learners worldwide to build GPT models from scratch through 8 hands-on video lectures covering everything from backpropagation to Transformer architecture and Byte Pair Encoding tokenization.
Anthropic signs a massive $45 billion deal with SpaceX, paying $1.25 billion per month through 2029 for AI computing power across two SpaceX data centers, including the Memphis-based Colossus 1, to fuel surging demand for its Claude AI software.
Google launches 'Gemini for Science,' a groundbreaking AI-powered suite partnering with 100+ institutions including Stanford and Imperial College London to revolutionize scientific discovery through hypothesis generation, computational analysis, and access to 30+ life science databases.
A new 1B parameter AI model called HRM-Text is making foundation model pretraining accessible to nearly anyone, requiring up to 900x less data and 600x less compute than traditional methods, bringing full pretraining costs down to roughly $1000 using a hierarchical recurrent architecture with strong benchmark performance across GSM8k, MATH, …
Nvidia is partnering with Ineffable Intelligence, the $1.1 billion-backed AI startup founded by DeepMind reinforcement learning pioneer David Silver, to co-develop AI systems that learn through trial and error using Nvidia's latest Grace Blackwell chips — pushing beyond conventional AI trained on human data toward systems that continuously discover new …
Adaption launches AutoScientist, a groundbreaking AI tool that simultaneously co-optimizes data and models through automated fine-tuning, claiming to more than double win rates across different models while making frontier-level AI training accessible beyond major labs.
Microsoft's AI-driven vulnerability system MDASH has discovered 16 previously unknown Windows flaws, including four critical remote code execution vulnerabilities now patched, using over 100 specialized AI agents to scan source code and top a major cybersecurity benchmark.