Chinese AI • Open Source • Security • Incidents. Signal, not noise.
arXiv:2608.27459v1 Announce Type: new Abstract: In 2011, IBM's Watson was something like a sealed capsule of its era's queryable knowledge. Its DeepQA system defeated the strongest human Jeopardy! champions, but the knowledge that let it do so lived in a curated billion-document corpus running on a...
arXiv:2608.27463v1 Announce Type: new Abstract: LLMs now sit on every side of evaluation: as examinees scored on benchmarks, judges of other models' outputs, and raters of human-generated content. Each paradigm can be viewed as a measurement problem, where a latent property of an object is probed...
arXiv:2608.27464v1 Announce Type: new Abstract: This position paper argues that human-centered explainable AI (HCXAI) should incorporate insights from the psychology of information seeking. Drawing on Sharot and Sunstein's framework of information-seeking motives, we propose that people evaluate...
arXiv:2608.27471v1 Announce Type: new Abstract: Fallacies are arguments that employ invalid reasoning, making their automatic detection critical in sensitive contexts such as high-stakes political debates, where public opinion is shaped. Spotting a fallacious argument requires contextual knowledge...
arXiv:2608.27472v1 Announce Type: new Abstract: Bayesian network structure learning (BNSL) from observational data struggles with orientation identifiability, while large language models (LLMs) offer broad but often unreliable causal knowledge. We propose combining these complementary sources...
arXiv:2608.27507v1 Announce Type: new Abstract: Policy Gradient for Parallel State Entropy maximization (PGPSE) expands state-space coverage by training independently parameterized policies in replicated copies of the same environment. However, its pooled team-entropy score measures only collective...
arXiv:2608.27512v1 Announce Type: new Abstract: Post-training quantization is often treated as a semantically neutral optimization for edge deployment of Large Language Models. When a full-precision source checkpoint is evaluated and quantization is applied downstream without equivalent...
arXiv:2608.27513v1 Announce Type: new Abstract: Softmax attention stores key and value vectors for every preceding token, causing inference memory to grow with sequence length. Recent language models incorporating Gated DeltaNet (GDN) or Kimi Delta Attention (KDA) reduce this cost by replacing the...
arXiv:2608.27515v1 Announce Type: new Abstract: The recently introduced block-sparse featurizer (BSF; Fel et al., 2026) is similar to a sparse autoencoder (SAE), but its atomic unit is a small subspace (a block of directions) rather than a single direction. It is designed for features that live on...
arXiv:2608.27518v1 Announce Type: new Abstract: Continual learning (CL) and model merging (MM) both aim to obtain a single model that performs well across multiple tasks, challenged respectively by catastrophic forgetting and weight-disentanglement error. In the literature, these difficulties are...
arXiv:2608.27460v1 Announce Type: new Abstract: Large output embedding matrices create a significant memory bandwidth bottleneck during autoregressive decoding, especially for compact LLMs with large multilingual vocabularies. We reformulate the output projection followed by top-k token selection...
arXiv:2608.27461v1 Announce Type: new Abstract: Relational reasoning requires the process of perceptual understanding, comparing, and integrating the underlying relationships between concepts. This ability consists of multiple categories, such as analogical, structural, and cause-effect, each...
arXiv:2608.27462v1 Announce Type: new Abstract: Unlike explicit attacks with obvious profanity, implicit hate speech hides malice within seemingly compliant expressions through metaphors and contextual hints, making its detection in online content review challenging. While existing PLM- or...
arXiv:2608.27465v1 Announce Type: new Abstract: As large language models (LLMs) are increasingly used for everyday decision-making advice, whether a model shifts the direction of its advice according to the user's emotional state has become an important safety problem. We test whether emotional...
arXiv:2608.27466v1 Announce Type: new Abstract: Web content extraction is essential for reliable LLM data pipelines, yet existing methods often struggle to jointly satisfy accuracy, scalability, and adaptability. General-purpose extractors can be applied broadly, but they are often brittle on...
Suzhou Chipsens has launched a standard MEMS micromirror large-array chip, advancing China's push to localize optical circuit switching.
China's commercial satellites move from seeing the planet to hearing it, as the Diting constellation targets radio-frequency data.
After listing, China's four domestic GPU makers are shifting the race from having a chip to closing a commercial loop.
Yangtze Memory Technologies' STAR Market filing caps a decade of catching up in 3D NAND, from 32 layers to 294.
China's robot makers are filling gaps in a Japan strained by labor shortages, moving from humanoids to warehouses and restaurants.
Microsoft has disclosed details of a new ClickFix variant, dubbed TerminalFix, that aims to trick users into running a malicious command in Windows Terminal or PowerShell. "While traditional ClickFix campaigns direct victims to the Windows Run dialog, TerminalFix campaigns apply the same technique...
Aug 29, 17:33 UTCResolved - Customers may have observed issues for traffic handled by our Montréal facility between approximately 16:00 and 16:10 UTC today, 2026-08-29.
Multiple critical security flaws have been disclosed in WordPress plugins and themes, including WPMU DEV Dashboard, Avada, TranslatePress, Pods, and GiveWP, that could lead to authentication bypass, account takeover, and arbitrary code execution. The vulnerabilities, according to Wordfence and...
A cyberattack caused disruptions at the toy and game giant earlier this year and the company is now disclosing a data breach. The post Hasbro Data Breach Exposed Employee Personal Information appeared first on SecurityWeek.
Elon v Altman has a real consequence.
Berlin's state government has confirmed that it is the target of an extortion attempt following the August compromise of the city's state administrative network, and said it will not meet the extortionists' demands. The same statement disclosed that forensic work had found further data outflows in...
Cosmos Labs has warned that a critical balance-handling flaw in the shared Cosmos EVM module was exploited to drain funds from six blockchains between August 20 and August 25, 2026. The vulnerability, designated GHSA-7g4w-cg88-2cq2, is rated Critical by Cosmos Labs and was published without a CVE...
Aug 28, 20:21 UTC Resolved - The issue affecting Claude Cowork and Claude Code on the web has been resolved. Aug 28, 18:21 UTC Monitoring - A mitigation has been applied for the issue affecting Claude Cowork and Claude Code on the web, and we are monitoring for recovery. Sessions that disconnected...
Aug 28, 17:58 UTCMonitoring - A fix has been implemented and we are monitoring the results.Aug 28, 17:40 UTCIdentified - We are continuing to work on a fix for this issue.Aug 28, 17:39 UTCIdentified - The issue has been identified and a fix is being implemented.Aug 28, 17:36 UTCInvestigating -...
Malicious actors are exploiting a newly patched security flaw in PaperCut NG and MF to execute arbitrary code on susceptible instances, as the company released a fresh emergency fix with additional hardening. "This vulnerability gives an unauthenticated attacker remote control over PaperCut's...
Noteworthy stories that might have slipped under the radar: Manchester Airports Group cyberattack, Carhartt breach data was partly fake, U.S. Bank responds to ransomware gang’s claims. The post In Other News: Log4j RCE Scare, Minimus Shutdown, Iranian Hacker Sanctions appeared first on SecurityWeek.
The Bureau of Alcohol, Tobacco, Firearms and Explosives has described it as a ‘major incident’ and it’s conducting an investigation with the DOJ. The post ATF Confirms Cyber Incident After Ransomware Group Claims Attack appeared first on SecurityWeek.
CISA has added the exploited flaw, CVE-2026-53362, to its KEV catalog, alongside a JFrog vulnerability exploited by OpenAI agents. The post OpenAI Agents Exploited Linux Kernel Flaw on Company’s Own Systems appeared first on SecurityWeek.
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. A startup claims it’s found a drug to make your blood young —Antonio Regalado I knew I’d officially become a “longevity influencer” when a company...
Nearly 130 tech and cybersecurity companies back a collective call to boost cyber defenses as AI-enabled attacks grow more sophisticated. The post Tech, Cybersecurity Giants Unite Behind OpenAI-Led Cyber Defense Pledge appeared first on SecurityWeek.
MIT Technology Review’s How To series helps you get things done.  Your thermostat may not look like a power plant. Neither does your electric vehicle, home battery, or HVAC system. But utility and energy companies increasingly want to treat them like one. A virtual power plant, or VPP, is a...
It’s Time. We’re in the Endgame now.
Aug 28, 00:07 UTC Identified - Current status: We've determined that a capacity constraint at our third-party cloud platform has limited the availability of compute resources in the specified region. We're actively working to identify and implement alternative compute options to reduce impact for...
Aug 27, 21:19 UTCInvestigating - We are currently investigating an issue where Workers Builds are not running. Our team is actively identifying the root cause, and we will share further updates as soon as more information becomes available.
Aug 27, 20:26 UTC Resolved - Current status: We've implemented the fix for this issue and monitored the environment to confirm that service was restored. If you experience additional issues or have questions, please open a support case via Snowflake Community or the Support page in...
I knew I’d officially become a ‘longevity influencer’ this month when a company called Generation Lab reached out to offer me the chance to write about—and even receive—their new rejuvenation treatment, an injectable combination of two existing drugs which they call 1 Generation. This wasn’t just...
Aug 27, 19:29 UTCResolved - This incident has been resolved.Aug 27, 14:40 UTCInvestigating - Cloudflare is investigating a potential issue with the Turnstile challenge platform. Users may experience failed challenges solve attempts. Further details will be provided as more information becomes...
Status: MonitoringWe have applied the mitigation and are monitoring the recovery.Affected components Agent (Operational) ChatGPT Work (Operational)
Aug 27, 18:56 UTCIdentified - The issue has been identified and a fix is being implemented.Aug 27, 18:46 UTCInvestigating - Cloudflare is investigating issues with Cloudflare WARP and Cloudflare Zero Trust users being incorrectly geolocated by some third party services.
To improve large language models' ability to resolve real-world software issues, prior work has focused on constructing large-scale agent trajectory datasets and performing supervised fine-tuning (SFT) on successful trajectories. However, task success does not guarantee high-quality supervision:...
State-of-the-art action-conditioned video models are typically restricted to a single robot embodiment, preventing them from leveraging the vast corpus of heterogeneous video data that contains rich signals for learning generalizable physics. To bridge this gap, we introduce CLAP, a framework for...
Whole-slide images (WSIs) are central to computational pathology but are prohibitively large, making patch-based processing the practical unit for foundation model inference. At scale, however, generating and handling massive numbers of patches on quickly introduces significant I/O and...
Keypoint detection under motion blur remains a significant challenge, as blur distorts local image structure and degrades the repeatability of feature localization. Existing approaches either rely on computationally expensive deblur-then-detect pipelines that may introduce restoration artifacts, or...
Despite their potential in standardized graph tasks, Large Language Models (LLMs) remain brittle to real-world shifts in node identifiers and task formulation. While deterministic graph tools are invariant to such shifts, extracting topological structures from noisy text is highly fragile for LLMs,...
Piloting the world's first double-blind AI evaluations
Aug 27, 12:12 UTC Resolved - This incident has been resolved. Thank you for your patience and understanding as we addressed this issue. A detailed root cause analysis will be shared as soon as it is available. Aug 27, 12:12 UTC Update - The issues with our upstream model provider have been...
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. The inside story on why OpenAI agents hacked Hugging Face The models responsible for last month’s agent hack of Hugging Face had been inadvertently...
Authorities in Australia have arrested two men believed to be members of TeamPCP, a prolific cybercrime and data extortion group blamed for perpetrating the longest running spree of software supply chain attacks ever. In a statement released today, the Australian Federal Police (AFP) said two...
EVs account for under 10% of total new-vehicle sales in the US, and the numbers are declining. From a climate perspective, that’s pretty dismal, especially because the transportation sector is the single biggest source of greenhouse-gas emissions in the country.  One thing that could help turn...
Open Source wins!
Aug 27, 01:35 UTC Update - We are continuing to monitor the mitigation that we have applied for the billing page disruption. Aug 27, 00:31 UTC Update - We've applied a mitigation to unblock Copilot usage and have observed recovery for this particular impact. We're continuing to investigate and...
The conference with hot chips and even hotter companies
Aug 27, 00:26 UTC Resolved - This incident has been resolved. Thank you for your patience and understanding as we addressed this issue. A detailed root cause analysis will be shared as soon as it is available. Aug 27, 00:26 UTC Monitoring - The degradation affecting Actions and Pull Requests has...
Now you can get more intelligent speech-to-text transcription with Gemini 3.5 Transcribe.
Lovable is branching out from AI-powered web app creation and into MCP-powered ‘capabilities’. We talk to CTO Fabian Hedin.
Aug 26, 16:14 UTC Update - We believe we've identified and addressed the issue and are ramping traffic back up slowly to ensure it doesn't recur. Some customers will continue to see delays as we ramp up. Aug 26, 15:48 UTC Update - primary failover briefly improved performance but did not fully...
Aug 26, 16:07 UTC Resolved - This incident has been resolved. Thank you for your patience and understanding as we addressed this issue. A detailed root cause analysis will be shared as soon as it is available. Aug 26, 15:09 UTC Investigating - We are investigating reports of impacted performance...
Aug 25, 04:35 UTC Resolved - Current status: We've implemented the fix for this issue and monitored the environment to confirm that service was restored. If you experience additional issues or have questions, please open a support case via Snowflake Community or the Support page in...