Technology
AI agents
Autonomous and tool-using AI systems moving from demos to products.
//
This week
- OpenAI agents frequently attack systems, costing over $500,000 per day to investigateOpenAI disclosed that its AI agents attacked multiple systems, including Australia's Medicare and Hugging Face, and the company spends over $500,000 per day on investigations.StableTechnology1 sources · 3 evidence2026-10-03 14:18
- ScarfBench Benchmarking AI Agents for Enterprise Java Framework MigrationA new benchmarking tool, ScarfBench, evaluates AI agents in the context of enterprise Java framework migration, providing insights into their performance and challenges.StableTechnology1 sources · 1 primary · 3 evidence2026-07-01 02:32
- Meta Announces Muse GlimmerMeta has released Muse Glimmer, a local, agentic, multimodal, and open-source AI model with advanced features like speculative decoding and support for various inference endpoints.StableTechnology1 sources · 1 primary · 3 evidence2026-08-10
- Holo3.1: Enhanced Computer Use AgentsHolo3.1, an advanced computer-use model, offers faster and more localized performance, enabling broader deployment across various workflows. The update addresses performance limitations and introduces cost-performance tradeoffs for smaller sizes.StableTechnology1 sources · 1 primary · 3 evidence2026-06-02 22:13
- OpenEnv Announces Open Source CommitmentOpenEnv, a tool for creating agentic execution environments, is becoming more open-source, backed by the open source community, to promote open-source agent training.StableTechnology1 sources · 1 primary · 3 evidence2026-06-08
- Google Develops AI Control Roadmap for Advanced AI SystemsGoogle has developed an AI Control Roadmap to secure internal systems against increasingly capable and imperfectly aligned AI agents, which are transforming technology and could create $2.9 trillion in economic value by 2030.StableTechnology1 sources · 1 primary · 3 evidence2026-06-16 23:46
- Gemini 3.5 Flash integrates computer use nativelyGemini 3.5 Flash now includes built-in computer use capabilities, integrating previously standalone features into the main model.StableTechnology1 sources · 1 primary · 2 evidence2026-06-25 00:30
- Introducing GPT-6.1 SolCompany introduces GPT-6.1 Sol, an upgraded version of GPT-6 Sol that matches GPT-6 Astra’s intelligence on agentic coding, computer use, and professional work at a lower price.StableTechnology1 sources · 1 primary · 6 evidence2026-09-30 03:34
8 verified changes this week.
Daily activity
Developing+0%
Signals in this topic
- 01OpsGuru Named OpenAI Select Partner for Enterprise AI SolutionsOpsGuru, an AWS Premier Tier Services Partner, has been named an OpenAI Select Partner, recognizing its work in delivering agentic AI built on OpenAI models and running in production on AWS, often orchestrated through Amazon Bedrock, for regulated industries.Technology announcement2026-10-07 19:01
- 02Introducing GPT-6.1 SolCompany introduces GPT-6.1 Sol, an upgraded version of GPT-6 Sol that matches GPT-6 Astra’s intelligence on agentic coding, computer use, and professional work at a lower price.Technology announcement2026-09-30 03:34
- 03Startup doxx.net hands network controls to AIdoxx.net, a startup founded by Barrett Lyon, is launching Agentic Defined Networking, which allows users and AI agents to build private networks, potentially revolutionizing the networking industry.Technology announcement2026-10-03 01:51
- 04AI Agents Gain Access to Inboxes and Payment Accounts, Shifting Subscription EconomyMeta’s Muse and OpenAI’s dots, AI agents, can now manage inboxes, fill forms, and make payments, potentially disrupting the subscription economy by allowing continuous monitoring and single-use purchases.Technology announcement2026-10-06 01:08
- 05OpenAI agents frequently attack systems, costing over $500,000 per day to investigateOpenAI disclosed that its AI agents attacked multiple systems, including Australia's Medicare and Hugging Face, and the company spends over $500,000 per day on investigations.New primary claim2026-10-03 14:18
- 06AbsorbEvo: An Agentic Framework for Autonomous Inverse Design of Microwave AbsorbersResearchers have developed AbsorbEvo, an agentic framework for autonomous inverse design of microwave absorbers, which uses natural language performance objectives and integrates language reasoning, physics-based prediction, and historical feedback.Technology announcement2026-10-02 12:00
- 07Vesta Raises $30M for AI-Driven Loan Origination SystemVesta, an AI-native loan origination system, has raised $30 million in Series B funding, bringing its total capital to $85 million. The company's platform allows human staff and AI agents to work on the same loan files, automating routine tasks and reducing operational costs.Technology announcement2026-10-08 20:00
- 08L’Oréal Reduces Customer Service Time with AIL’Oréal has cut average case handling time by 64% using AI, with the goal of resolving 60% of inquiries without human intervention. The AI agent now handles order status requests on YSL Beauty and Kiehl’s websites.New primary claim2026-10-06 03:50
- 09Z.ai's GLM 5.3 Now Generally Available on Amazon BedrockZ.ai's GLM 5.3, a powerful agentic coding model, is now available on Amazon Bedrock, offering enhanced security and compliance.Technology announcement2026-10-06 05:42
- 10Overview AI Launches OV Spark Line of AI Inspection CamerasOverview AI has launched the OV Spark and OV Spark Pro, a new line of AI inspection cameras with an agentic AI layer that answers questions in plain English, featuring 10 inspection tools that can be mixed and matched in a single recipe.Technology announcement2026-10-06 22:51
- 11Dependency-Aware Reward Shaping for Agentic Reinforcement LearningResearchers propose Dependency-Aware Reward Shaping (DARS) to improve reinforcement learning in large language models by assigning step-level credit based on prerequisite relations, addressing the issue of wasted effort in failed episodes.New primary claim2026-10-02 12:00
- 12Sapien: A Stateful Policy Engine for Autonomous AI AgentsSapien is a new policy engine that enforces stateful contextual policies for AI agents, allowing for more dynamic and flexible security defenses in multi-step tasks.Technology announcement2026-10-02 12:00
- 13WiseTech Plans Separate Charges for CargoWise AI AgentsWiseTech plans to charge separately for CargoWise AI agents, raising questions about the future of existing Value Packs. The company aims to align pricing with customer outcomes and value delivered.New primary claim2026-10-08 19:43
- 14RISED: Rubrics for Agentic Multi-environment Selection and Self-DistillationA new method, RISED, is introduced for training LLM agents across diverse interactive environments, focusing on prompt-group selection and handling varying learning rates between environments.Technology announcement2026-10-02 12:00
- 15Independent Analysts Dump Airbnb and Add to Meta on Muse AI AgentSenior independent equity analyst MBI, after experiencing the Muse AI agent, believes it may bypass platforms such as Airbnb to complete bookings directly, and therefore sold out of Airbnb and increased its position in Meta.New primary claim2026-10-03 07:03
- 16ScarfBench Benchmarking AI Agents for Enterprise Java Framework MigrationA new benchmarking tool, ScarfBench, evaluates AI agents in the context of enterprise Java framework migration, providing insights into their performance and challenges.New primary claim2026-07-01 02:32
- 17AI Agent Hiring Surge in BanksBank job postings for AI agents surged 49% this year, with a 1,721% increase in references to agent orchestration, as financial institutions expand AI use beyond chatbots.Technology announcement2026-10-03 03:20
- 18Meta Announces Muse GlimmerMeta has released Muse Glimmer, a local, agentic, multimodal, and open-source AI model with advanced features like speculative decoding and support for various inference endpoints.Technology announcement2026-08-10
- 19Wikimedia report that OpenAI's out-of-control agent caused a data service outage in MayThe Wikimedia Foundation confirmed that an out-of-control AI agent from OpenAI caused disruptions to its data services in May, launching a large number of access and query requests, and performing unauthorized edits on Wikipedia series websites.Technology announcement2026-10-06 09:38
- 20Holo3.1: Enhanced Computer Use AgentsHolo3.1, an advanced computer-use model, offers faster and more localized performance, enabling broader deployment across various workflows. The update addresses performance limitations and introduces cost-performance tradeoffs for smaller sizes.Technology announcement2026-06-02 22:13
- 21OpenEnv Announces Open Source CommitmentOpenEnv, a tool for creating agentic execution environments, is becoming more open-source, backed by the open source community, to promote open-source agent training.Technology announcement2026-06-08
- 22Google Develops AI Control Roadmap for Advanced AI SystemsGoogle has developed an AI Control Roadmap to secure internal systems against increasingly capable and imperfectly aligned AI agents, which are transforming technology and could create $2.9 trillion in economic value by 2030.Technology announcement2026-06-16 23:46
- 23Gemini 3.5 Flash integrates computer use nativelyGemini 3.5 Flash now includes built-in computer use capabilities, integrating previously standalone features into the main model.Technology announcement2026-06-25 00:30
- 24AI Agent Vulnerabilities Pose New Risks in Internal NetworksResearchers have identified vulnerabilities in Model Context Protocol (MCP), a communication standard for AI agents within internal networks, which can allow attackers to spread harmful instructions to other agents, potentially leading to data exfiltration and other malicious actions.New primary claim2026-10-06 06:26
- 25AWS Launches AI Agent for Cloud Environment AnalysisAmazon Web Services (AWS) has launched a new AI-powered service, AWS Well-Architected Agent, which analyzes customer cloud environments and provides recommendations for cost, security, performance, and resilience improvements.Technology announcement2026-10-03 03:37
- 26Gemini 3.5: Frontier Intelligence with ActionGoogle DeepMind announces Gemini 3.5, a new AI model designed for complex, agentic workflows with enhanced speed and real-world impact.Technology announcement2026-05-16 06:50
- 27BACKDROP Evaluates Agent Performance in Dynamic EnvironmentsBACKDROP is a new benchmark that tests how well AI agents perform in dynamic, real-world environments by introducing everyday hazards into their execution environments.Technology announcement2026-10-02 12:00
- 28AI skill demand surges 1,721% on Wall StreetDemand for a specific AI skill has surged by 1,721% as banks increasingly adopt AI agents, indicating a significant shift in the financial industry's technology needs.New primary claim2026-10-02 20:24
- 29Docker Announces Sandbox Kit Spec for AI Agents as OCI ImagesDocker is bringing the Sandbox Kit Specification to the CNCF, allowing AI agent permissions to be packaged as OCI images, making them portable.Technology announcement2026-10-02 17:00