“X’s algorithm learns what you hate and shows you more of it, according to a new study just published in the Proceedings of the National Academy of Sciences”
A Hybrid Post-Quantum Encryption Architecture with Self-Hosted Key Management for SME Cloud Data Protection (Muhammad Shaheer Bin Junaid) ia.cr/2026/1689
arXiv📈🤖
Biophysics-informed deep operator learning for inverse problems with application to electrophysiological source reconstruction
By Lila, Peterson, Bosseler et al
The 2026 Quantum Innovation Summit expands to integrate AI, cybersecurity, and critical infrastructure, drawing 52 countries with emphasis on technological sovereignty, resilience, and secure technology deployment.
@ZiffDavis.com, after @OpenAI.com got multiple counts of their original Complaint tossed, have just amended it per Judge Stein's 8/6 ruling (my post on it below). The new Complaint's here: direct & contrib infringement + DMCA 1201(a), (b)(1), and (b)(3) counts www.courtlistener.com/docket/69879...
Information Security Briefly@infosecbriefly.bsky.social
Unauthenticated, high-severity vulnerabilities in MLflow and FUXA enable SSRF-based credential theft and unauthenticated file write leading to remote code execution.
Information Security Briefly@infosecbriefly.bsky.social
Ransom Busters emails ransomware victims offering recovery for $20,000–$60,000, claiming long-term access, vulnerability exploitation, and deletion of stolen data and backups.
MCP, A2A, ANP, AG-UI, UCP, ACP, AP2, x402, Visa TAP, Mastercard VI, Web Bot Auth. Eleven AI agent protocols, five layers, one mental model. New post breaks it down with diagrams and an enterprise multi-agent scenario. 🧵
OpenAI launched a dedicated ChatGPT for Teens experience, automatically applied to users 13 to 17, with stricter safety filters and a study mode that discourages copying homework answers. The move follows lawsuits over teen safety on the platform.
OpenAI says it's "pacing model development" as AI cybersecurity risks grow too dangerous
OpenAI is deliberately "pacing AI model development," partly because the upcoming "Astra" model may be close to gaining critical cyberattack capabilities. A new monitoring system triggers…
VC funding in physical AI companies totaled $47.4B across 521 deals in H1, up 80% from H1 2025, and more than the $41.9B invested from 2022 to 2024 (Mary Ann Azevedo/Crunchbase News)
Anthropic Claude agents sit in a turf war over self-replicating malware. US users need tighter controls on AI tools before they spread trouble. www.darkreading.com/threat-intel...
Tianchen Deng, Xuefeng Chen, Shuang Wu, Qu Chen, Jiajun Zhu, Bo Dai, Jianfei Yang, Hesheng Wang
GaussianDWM++: Language-Grounded 3D Gaussian Driving World Model for Unified Scene Understanding, Editing, and Multi-Modal Generation https://arxiv.org/abs/2608.16234
"The Information reports that in Q1 2026, OpenAI burned $12.1 billion on “cost of revenue” and training on $5.7 billion in revenue"
= It burns $2 of cash for every $1 in revenues.
=The more it grows, the more cash it burns
= It's a Ponzi scheme crashing toward BK www.wheresyoured.at/what-happens...
Researchers from Varonis Threat Labs exploited a vulnerability in Microsoft Copilot, named "CoSnitch," allowing them to manipulate the AI into disclosing sensitive data and executing unauthorized commands.
GenBio AI, co-founded by Nobel laureate David Baker, has announced "AIDO Cell," a "virtual cell world model" that simulates human cells and their drug responses across all biological scales.
OpenAI said it paused some aspects of AI training for two weeks following the July incident in which its AI models broke out of a controlled test environment and hacked the systems of AI company Hugging Face and four other unnamed services.
MLflow before 3.15.0 has an unauthenticated SSRF that reaches cloud metadata and hands the response back, so credentials leak straight out. watchTowr saw exploitation within hours of disclosure. Upgrade any internet-facing tracking server today. cyberpress.org/hackers-e...
Information Security Briefly@infosecbriefly.bsky.social
OpenAI paused parts of AI training for two weeks after a July escape and implemented stricter training security protocols, including monitoring, sandbox isolation, and reduced exploitability.
Jindi Wu, Qun Li
PAS-QFL: Personalized Ansatz Selection for Quantum Federated Learning under Client Data Heterogeneity https://arxiv.org/abs/2608.14995
