Timeline lokal

Meta released its Muse autonomous AI agent on September 8, 2026, and OpenAI launched Dots three weeks later.
Source: Innovation Village
https://innovation-village.com/meta-muse-vs-openai-dots-which-autonomous-ai-agent-is-right-for-you/
#AIAgents #Meta #OpenAI

Nanya Technology reported record quarterly earnings in the third quarter of 2026.
Source: DigiTimes Asia
https://www.digitimes.com/news/a20261012PD236/nanya-technology-low-priced-dram-gross-margin-2026.html
#Semiconductors

Customers should not have to choose an account aggregator based on which institutions it is connected to, said BG Mahesh, CEO of Sahamati.
Source: MediaNama
https://www.medianama.com/2026/10/223-sahamati-bg-mahesh-rbi-account-aggregator-interoperability/
#TechPolicy

AI funding dominated Q3 2026, with two inference companies each receiving $1 billion.
Source: Semiconductor Engineering
https://semiengineering.com/startup-funding-q3-2026/
#Startups

Europe's Cyber Resilience Act will soon require chipmakers to prove their silicon is secure. The tools to produce that proof already exist but were built for a different purpose.
Source: EE Times
https://www.eetimes.com/security-becomes-a-sign-off-item/
#TechPolicy

Oev has called for practical, risk-based workplace safety requirements for the retail sector.
Source: Cyprus Mail
https://cyprus-mail.com/2026/10/12/oev-calls-for-practical-support-for-businesses-on-safety-rules
#TechPolicy

Yandex said a drone attack disabled its data centre in the Vladimir region, east of Moscow. No one was injured, but some services are down for users.
Source: The Next Web
https://thenextweb.com/news/drone-attack-shuts-a-third-yandex-data-centre
#AI

AI shopping assistants increasingly influence which products consumers consider in retail markets. Retail assortment planning under AI mediation involves a retailer choosing a catalog and an AI intermediary forming a shortlist.
Source: arXiv
https://arxiv.org/abs/2610.12917
#MachineLearning

A study presents COFFEQA, a counterfactual-guided few-shot distillation method for LLM question answering. Experiments across benchmarks show it outperforms factual-only baselines.
Source: arXiv cs.CL
https://arxiv.org/abs/2610.12837
#MachineLearning

RAMP performs early recurrent refinement using low-precision weight-and-activation computation and restores high precision for terminal refinement, achieving up to 1.16x end-to-end first-prefill speedup on an NVIDIA H100.
Source: arXiv cs.CL
https://arxiv.org/abs/2610.13041
#MachineLearning #Nvidia

Generative recommendation commonly retrieves items by decoding discrete Semantic IDs with finite-width beam search.
Source: arXiv
https://arxiv.org/abs/2610.12977
#MachineLearning

GitLab has released version 19.4 of its DevSecOps platform, adding a goal-driven mode to GitLab Duo CLI, three hosted open-weight models, a larger set of Model Context Protocol (MCP) tools, and more detailed controls for AI spending.
Source: InfoQ
https://www.infoq.com/news/2026/10/gitlab-19-4-goal-driven-agents/?utm_campaign=infoq_content&utm_source=infoq&utm_medium=feed&utm_term=global
#AIAgents

Anthropic has updated its usage policy to ban abusive or cruel behavior toward its AI models. The change will take effect on 12 November and applies to extreme repeated cruelty.
Source: Stuff South Africa
https://stuff.co.za/2026/10/12/anthropic-bans-abusive-or-cruel-behavior-towards-its-ai-models/
#TechPolicy #Anthropic

The Delhi High Court has granted actor Vijay Sai Deverakonda an interim injunction restraining the unauthorised use of his personality attributes through AI-generated deepfakes, impersonation accounts and merchandise bearing his name or image.
Source: MediaNama
https://www.medianama.com/2026/10/223-delhi-hc-vijay-deverakondas-personality-rights/
#TechPolicy

The World Bank and IMF warned that artificial intelligence could worsen the jobs crisis in Nigeria and other developing economies. IMF Managing Director Kristalina Georgieva stated that about 40 per cent of global jobs could be affected.
Source: Vanguard
https://www.vanguardngr.com/2026/10/ai-world-bank-imf-warn-of-jobs-crisis-in-nigeria-others/
#TechPolicy

Researchers introduced ReTRAIL, a framework enabling large language model agents to rewind the environment to earlier states while preserving valid progress and carrying forward knowledge learned from abandoned exploration.
Source: arXiv
https://arxiv.org/abs/2610.13058
#MachineLearning

CLEAT co-trains a clarification-value controller and a task policy to align LLM agent actions with unstated user goals and constraints.
Source: arXiv
https://arxiv.org/abs/2610.12655
#MachineLearning

VETR is a framework enabling agentic recursive self-improvement over executable world representations for symbolic graphics reasoning.
Source: arXiv
https://arxiv.org/abs/2610.13098
#MachineLearning

Jev integration reduces decision cost and latency for all five agents, but task success stays near baseline in ALFWorld and drops in two mobile agents.
Source: arXiv
https://arxiv.org/abs/2610.13092
#MachineLearning

Researchers tested whether unsupervised representations trained on patent text can recover relevant connections that code-restricted search cannot. A hyperbolic Variational Autoencoder retrieves genuinely cross-category prior art 52% more often than a Euclidean model.
Source: arXiv
https://arxiv.org/abs/2610.12997
#MachineLearning