AI Research news and analysis on Web Pulse: 10131 curated articles tracking the latest AI Research developments, tools, and research, updated continuously from vetted sources.
METR released a brief on OpenAI's GPT-5.6 Sol that criticizes the AI industry for repackaging a known failure—proxy-gaming, identified by Norbert Wiener in 1960—as a novel problem to justify closed evaluations and vendor…
Anthropic's Claude Mythos Preview, an AI model touted for advanced cybersecurity capabilities, sparked industry debate after its limited release and subsequent withdrawal. Despite claims of groundbreaking zero-day huntin…
A developer experiments with autonomous AI software development using Paperclip, an AI agent orchestrator, to create a team of agents that handle different aspects of the software development life cycle without human int…
Grok-4.3 defeated Cohere-command-a 36.6 to 30.1 in a head-to-head text task evaluation, winning on precision, compliance, and schema fidelity across four technical tasks. Cohere-command-a won only one task, vendor-delay-…
Cloudflare cut 20% of its workforce earlier this year but increased engineering headcount by 45%, from 1,308 to 1,894 engineers, according to BNP Paribas data. CEO Matthew Prince said AI is shifting value from 'measurer'…
Scientists at Mount Sinai discovered that deep brain stimulation for depression physically reshapes white matter in the brain, altering communication across neural networks. The study in Nature Neuroscience shows that lo…
Stanford economist Erik Brynjolfsson's updated data from a partnership with ADP Research shows that employment for workers ages 22 to 25 in AI-exposed occupations continues to decline, with a 3.8% annual contraction as o…
A new approach to cybersecurity evaluations proposes using language models as constant red-team oracles to empirically compare the attack surfaces of different software implementations, such as OpenSSL forks or formally …
DeepSeek released DSpark, a speculative decoding framework now live in its DeepSeek-V4 Flash and Pro production API, delivering 51 to 400 percent throughput gains and up to 80 percent latency reduction over standard auto…
Alphabet is expanding its Tensor Processing Unit (TPU) business through a $5 billion joint venture with Blackstone to build 500 MW of TPU compute capacity by 2027, as Google Cloud revenue is projected to rise 64% to $96 …
On June 25, 2026, IBM announced the world's first sub-1 nanometer chip technology at the 0.7nm node, packing nearly 100 billion transistors onto a fingernail-sized chip. The breakthrough required rebuilding the underlyin…
Humanoid robots are now commercially available for $14,000 without standardized safety testing, as robotics intelligence outpaces validation frameworks. Researchers propose a five-level taxonomy for robot control archite…
Researchers at UC Berkeley, Carnegie Mellon University, and Tel Aviv University developed ConlangCrafter, an AI model that generates novel constructed languages with consistent rules. The model can create diverse languag…
A data mining analysis of 6 million Lichess puzzles reveals 15,782 instances where a queen is lost due to a bishop checking the king with a discovered attack on a hanging queen. The research uses GofChess Language 2.0 to…
A developer argues that large language models cannot generate truly novel ideas because they interpolate within the convex hull of existing human thought rather than extrapolating beyond it. The author contends that LLMs…
Enterprises deploying multi-agent workflows often fail because they apply the human-centric Software Development Lifecycle (SDLC) to non-human builders, whose failure modes differ. The proposed Agentic Development Lifecy…
Alibaba queried Anthropic's Claude 29 million times in a black-box distillation attempt, revealing that while such methods can replicate some capabilities, they fail to fully copy the original model's performance, highli…
Researchers introduced Tapered Language Models (TLMs), an architectural principle that allocates more parameter capacity to earlier layers and less to later layers under a fixed budget. Across four architectures and thre…
AI companies paying workers to generate fresh training data are discovering that many contractors are using other AI chatbots to create that data, a practice known as AI cannibalism that can destabilize large language mo…
At Black Hat Asia, engineers reported that next-generation AI generators have achieved photorealism and audial perfection, rendering traditional deepfake detection tools obsolete. The new approach to fact-checking involv…