Timeline lokal

Researchers introduce Alignment Forecasting to predict alignment failures before training.
Source: arXiv cs.CL
https://arxiv.org/abs/2609.35805
#MachineLearning

LLM agents can make unsafe tool calls even when instructed to behave safely.
Source: arXiv cs.CL
https://arxiv.org/abs/2609.35807
#MachineLearning

Mnemon uses raw, dated records and divides memory into fast System 1 and slow System 2 work. It scores 91.7% on LoCoMo and 83.8% on LongMemEval-S using gpt-4.1-mini.
Source: arXiv cs.CL
https://arxiv.org/abs/2609.36059
#MachineLearning

LAUNCHub Ventures announced the first close of Fund III, securing €65 million with a target of over €75 million.
Source: EU-Startups
https://www.eu-startups.com/2026/10/sofia-based-launchub-ventures-secures-e65-million-first-close-of-fund-iii-targets-e75-million/
#Startups

ElevenLabs, an AI voice start-up, is valued at $22bn after a tender offer. The company says its AI agents handle more than 15 million customer service conversations each week.
Source: Silicon Republic
https://www.siliconrepublic.com/start-ups/ai-voice-start-up-elevenlabs-valued-at-22bn-after-tender-offer
#Startups

Huawei partnered with DeepSeek to launch an open-source programming software toolkit optimised for Ascend chips. DeepSeek stated the toolkit is free and includes the TileLang platform.
Source: TahawulTech
https://www.tahawultech.com/home-slide/huawei-collaborates-with-deepseek-on-ai-chip-software/
#Semiconductors #Huawei #DeepSeek

US Defense Secretary Pete Hegseth announced the establishment of the Autonomous Warfare Command.
Source: The Next Web
https://thenextweb.com/news/pentagon-autonomous-warfare-command-autowarcom
#TechPolicy

VisKG is a reinforcement learning framework that translates visual content into question-specific knowledge graph representations.
Source: arXiv cs.CL
https://arxiv.org/abs/2609.35942
#MachineLearning

More than 44,000 people have filed legal objections to their health data being handled by the Palantir NHS platform.
Source: The Guardian
https://www.theguardian.com/technology/2026/sep/30/legal-objections-palantir-nhs-platform-handling-data
#TechPolicy

A new paper introduces Hyperspherical Semantic Trajectory Analysis to track technology diffusion from text streams.
Source: arXiv cs.CL
https://arxiv.org/abs/2609.35845
#MachineLearning

Maybank Asset Management Singapore is widening access to the tokenised share class of its Singapore dollar money market fund through Synthesys Network.
Source: Fintech News Singapore
https://fintechnews.sg/138369/wealthtech/maybank-tokenised-sgd-money-market-fund-synthesys/
#Tech

BMW will cut 20% of senior management roles and use AI to streamline its leadership structure.
Source: Business Today
https://www.businesstoday.in/jobs/story/bmw-to-cut-20-of-senior-management-roles-with-ai-as-it-pushes-for-cost-savings-559029-2026-10-01?utm_source=rssfeed
#EnterpriseAI

CruxBench evaluates information discovery in large language models by grading generated questions using Value of Information. Frontier models only narrowly outperform a random-timing baseline.
Source: arXiv cs.CL
https://arxiv.org/abs/2609.35879
#MachineLearning

Lookahead-R is proposed as a planning-based framework that reformulates tool retrieval as a resource-constrained sequential decision-making problem.
Source: arXiv cs.CL
https://arxiv.org/abs/2609.35811
#MachineLearning

OpenAI said it identified and mitigated a coordinated model-distillation campaign involving individuals linked to Moonshot AI, the Chinese developer of Kimi models.
Source: The Information
https://www.theinformation.com/briefings/openai-accuses-moonshot-distillation-campaign
#AI #OpenAI

Researchers introduced SCOUT, a two-stage agentic safety verifier for computer-use agents that synergizes reasoning-intensive rubric generation with tool-intensive evidence gathering.
Source: arXiv cs.CL
https://arxiv.org/abs/2609.36201
#MachineLearning

A new study evaluates Large Language Models including Llama-3 8B, Qwen-2.5 14B, and Llama-3.3 70B for extracting specific attributes from SEC 10-K text.
Source: arXiv cs.CL
https://arxiv.org/abs/2609.35864
#MachineLearning #SEC #Llama

TypeSafe AI has released Jev, a decision-only model that returns typed probabilities instead of text.
Source: InfoQ
https://www.infoq.com/news/2026/10/typesafe-ai-jev-released/?utm_campaign=infoq_content&utm_source=infoq&utm_medium=feed&utm_term=global
#Tech

Google is withholding its Gemini 4 Argon artificial intelligence model from the public, releasing it only to a vetted group of cybersecurity experts.
Source: Malay Mail
https://www.malaymail.com/news/tech-gadgets/2026/10/01/google-holds-back-gemini-4-argon-limits-release-to-vetted-cybersecurity-experts/237250
#AI #Google #Gemini

California enacted new laws targeting companies using artificial intelligence to initiate job cuts or monitor employees.
Source: Engadget
https://www.engadget.com/2274224/california-law-bans-companies-from-relying-on-ai-to-fire-workers/
#TechPolicy