Topic
Explore English articles about Generative AI.
X identified a roughly 200,000-account "bot farm" suspected of ties to China. Only 200 of those accounts posted about AI and power policy, underscoring the need to assess scale and actual impact separately.
Qualcomm's High Bandwidth Compute technology is reportedly gaining support from Samsung and SK hynix. The design, which moves computation closer to memory, will be tested with AI250 samples and third-party verification in mid-2027.
C.L.A.Y. connects a Qiskit simulator, not actual quantum hardware, to Unity, using it to generate maps and branch the narrative. The game isolates this untested technology within a commercial RPG, and its real limits remain unverified.
Facing a worldwide RAM shortage, Google is tightening rules across Android 17 and Google Play, and starting in February 2027 will limit the store visibility and publication of memory-wasting apps. Here's a look at the new optimization push aimed at breaking Android's dependence on ever-improving hardware.
Alibaba has previewed its design for Qwen4 through Qwen3.8-Flash-Next. Despite having a capacity equivalent to 176 billion parameters, it activates only 6 billion, redesigning long-context inference and training efficiency.
The anonymous model Ox Alpha turns out to be Z.ai's GLM-5.3-Flash. Here's how its inference was designed to run on tens of thousands of Chinese-made AI chips, and why the claimed 100 trillion tokens-per-day capacity figure needs context.
TamperBench, presented at KDD '26 by an international research team including the University of Waterloo, shows that safety mechanisms in all 21 open-weight LLMs tested could be neutralized. The findings expose the limits of existing defense techniques and raise new questions for public procurement and safety evaluation.
SiFive's BigSky packs a 32-core RVA23-compliant RISC-V system into a 2U rack unit, aiming to accelerate OS and CUDA porting work. With limited availability, pricing, power consumption, real-world performance, and RHEL support status remain open questions.
Intel's Crescent Island pairs 32 Xe cores with up to 480GB of LPDDR5X in a 350W air-cooled PCIe card aimed at long-context inference. Memory bandwidth and real-world performance remain undisclosed.
Apple's M6 uses a 2nm process and three types of CPU cores to update its smaller Macs, while the M5 Ultra targets large-scale workloads with four third-generation 3nm dies and 1.2TB/s of memory bandwidth. We compare the official performance figures against previous generations and lay out the design differences and caveats.
