AI News Archive: August 4, 2026 — Part 12
Sourced from 500+ daily AI sources, scored by relevance.
- Cursor Open-Sources Mixture-of-Kittens (MoK): A Deterministic MoE Training Megakernel for GB300 NVL72 Racks
Cursor Open-Sources Mixture-of-Kittens (MoK): A Deterministic MoE Training Megakernel for GB300 NVL72 Racks MarkTechPost
- Open Secure AI Alliance proposes SAFE guidelines as membership tops 120
The Open Secure AI Alliance today proposed a set of guidelines for reporting cybersecurity incidents involving artificial intelligence agents, one week after the group was formed. The proposal is called Shared AI Findings Exchange, or SAFE, and was published as a request for comments by the Linux Foundation. Nvidia Corp., Cisco Systems Inc., CrowdStrike Holdings […] The post Open Secure AI Alliance proposes SAFE guidelines as membership tops 120 appeared first on SiliconANGLE .
- Open Secure AI Alliance Expands at Black Hat: What You Should Know
The Open Secure AI Alliance introduced SAFE guidelines and open agent-security tools at Black Hat, giving enterprises a framework for safer AI deployment. The post Open Secure AI Alliance Expands at Black Hat: What You Should Know appeared first on TechRepublic .
- AWS’s Kiro Crew aims to turn AI coding agents into autonomous engineering teams
AWS’s Kiro Crew aims to turn AI coding agents into autonomous engineering teams InfoWorld
- AI agents get better at IT ops, but only with humans in the loop
AI agents are performing roughly 1 in 3 actions in enterprise IT workflows (but that share is rising quickly), while human analysts are rejecting about one-quarter of AI-proposed actions (but that rate is falling), according to a new study of tens of thousands of human-AI interactions. Operational data, rather than underlying AI infrastructures, is often the culprit when things go wrong. Human analysts are approving the most consequential actions, managing exceptions, and supervising and shaping agentic systems, while AI agents are carrying out routine tasks and executions, automation platform provider Fixify found in the study . “That may sound less dramatic than replacing the help desk,” Matt Peters , Fixify’s co-founder and CEO, wrote in a blog post . “It’s also a much more credible path to changing how IT work gets done.” Building scaffolding Fixify identified four steps of agentic work: Planning, proposing, approving or declining, then acting on approved steps. It analyzed nearly 18,000 plans and over 147,000 actions executed by agents across 40 companies over a three-month period, finding that agents are taking over one-third of IT actions, most notably in software, applications, security, and collaboration work where requests tend to be “repeatable and easy to reverse.” Tasks that are well understood and that present low risk are best suited for the current generation of agents, Peters wrote. Human analysts remain closely involved in higher-stakes areas like identity verification, setting up and removing IT access (onboarding and offboarding), and hardware environments. However, AI’s share of the work is increasing as feedback loops improve: Over the three-month period, human approval of AI-proposed actions rose from 23% to 41%, and rejection fell from 27% to 16%, Fixify found. The company identified six types of actions in AI automation. Running a skill — actually doing something — accounted for 39.4% of all actions). Most of the rest were coordination: sending a message to the human requester (27.7% of actions), leaving an initial comment (13.2%), giving instructions to a human analyst (9.8%), or waiting (8.8%). Running entire workflows accounted for just 1.1% of actions. AI is building “scaffolding” that wraps around meaningful changes, often planning far more scenarios than the agent will execute. Typically, agents map out 15 possible actions but run only two, Fixify said. “The agent maps the paths a request could take, then walks down the path that makes the most sense as it meets reality,” the study said. Peters pointed to one example where an AI agent identified which team needed access to process a high-volume type of ticket. Rather than fully automating the process, the agent did the initial triage, asked questions, then routed tickets to the team that had the information to act immediately. “We didn’t need a world-ending hive mind,” he said. “We just needed to point a little conversational intelligence in the right direction.” When AI breaks down IT automation typically involves analyzing tickets and moving them along; in other words, low-risk tasks. But agents do participate in areas like security (albeit only about 6%), most notably adding and removing people from groups or channels, unlocking accounts, resetting passwords, analyzing multi-factor authentication (MFA), provisioning (or deprovisioning) accounts, and assigning software licenses. However, this identity-lifecycle work is where agents failed the most, particularly in onboarding and offboarding and identity-access management (IAM), the study found. “Hardware and connectivity changes rarely fail; identity-lifecycle changes fail three-to-nine times as often.” Why AI breaks down Thanks to human-in-the-loop controls, Fixify was able to analyze scenarios where agent recommendation diverged from human judgment. This occurred about 23% of the time. The largest failure category (nearly 50%) was ‘target not found,’ meaning the agent couldn’t uncover what it needed. This typically comes down to poor data: A user, group, account, or resource was not where the system expected it to be. When people change teams, groups are restructured, accounts are renamed, or work has already been done but not reflected in the system, this is more of an identity hygiene problem than an AI problem. The system needs cleaner and more current data. Invalid inputs accounted for around 29% of failures, followed by unhandled errors, denied permissions, or invalid operations or configurations. The latter signal “real breakage” in integrations, according to Fixify. AI becomes more sophisticated over time The good news is that AI automation improves over time, even if it might take a while. In hybrid systems , humans keep the most consequential changes under their own control, and iterative rejection and approval helps AI learn. Over time, agents’ plans get leaner and they start to re-plan when conditions change, rather than pre-planning all kinds of scenarios that may never occur. “That’s a sign of sophistication,” the study said. “Adapting in the moment is a more advanced behavior than trying to pre-script every contingency.” In turn, humans second guess the system less often and feel comfortable handing off more work. Instead, they control how agents behave, make high-impact decisions, and handle exceptions. “The hardest requests remain human-heavy, especially those that require repeated replanning or contextual judgment,” the study said. How teams can adapt to AI agents As agentic AI becomes embedded in more workflows — and at deeper levels — enterprises must evolve to accommodate, Fixify emphasized. This means investing in clean identity data and building strong playbooks, review workflows, and reliable integrations. Teams should judge agentic tools by their supervision loop and view rejections as a training process, Fixify advised. Analyst time, queues, and metrics should be built around reviewing proposals. Agent replanning can be seen as a routing signal: A single replan might indicate healthy adaptation, while repeated replanning means ambiguity, irrelevance, or unclear policies. “Make the review surface easy to understand so analysts can assess proposed actions and make quick decisions about how to proceed,” the study advised. “This is where the analyst’s attention belongs.”
- Vertiv Expands Global Manufacturing Capacity for AI-Ready Data Center Cooling Solutions
Vertiv announced investments at its Tognana campus near Padua, Italy, to expand manufacturing and integrated testing capabilities for data center cooling systems. The company expects the investments to double chiller production capacity in the region by the end of 2026 and plans to complete a new large-scale testing laboratory in early 2027, supporting growing demand for […] The post Vertiv Expands Global Manufacturing Capacity for AI-Ready Data Center Cooling Solutions appeared first on CXOToday.com .
- The Agent Development Lifecycle has arrived on Cloudflare
Agents can write code faster than teams can review, deploy, and maintain it. Today we’re introducing the Agent Development Lifecycle and the Cloudflare primitives that underpin it
- China draws up safety rules for autonomous vehicles
China has drawn up new safety rules for the growing number of autonomous vehicles on its roads, including a mandatory deactivation override, the government said Tuesday.
- Adnoc and SLB to use AI to cut workload across 120 rigs
Adnoc and SLB to use AI to cut workload across 120 rigs thenationalnews.com
- What Happens When AI Clones Your Face? Inside the Terrifying Rise of Digital Doppelgängers
What Happens When AI Clones Your Face? Inside the Terrifying Rise of Digital Doppelgängers PCMag Australia
- AMD’s AI engine shifts into higher gear as data center revenue more than doubles, Helios ramps & market is confused
Advanced Micro Devices Inc. delivered another strong quarter Tuesday, beating Wall Street expectations as its artificial intelligence infrastructure business continued its rapid expansion. The company posted 107% year-over-year growth in data center revenue, reinforcing that AMD is becoming a formidable challenger in the race to power enterprise AI. The Santa Clara, California-based chipmaker reported second-quarter […] The post AMD’s AI engine shifts into higher gear as data center revenue more than doubles, Helios ramps & market is confused appeared first on SiliconANGLE .
- Red Hat leads open-source project to automate AI governance
IBM Corp.’s Red Hat subsidiary today announced the formation of asago, an open-source community project intended to turn artificial intelligence governance policies into operational controls that can be deployed with AI systems. Short for AI Safety and Governance Orchestration, asago is intended to connect the work that is now often divided among compliance teams, data […] The post Red Hat leads open-source project to automate AI governance appeared first on SiliconANGLE .
- The Sovereign AI Choice: Why Enterprises Need to Act Now
As agentic AI reshapes how enterprises operate, the businesses pulling ahead may become the ones building sovereign control into their data and AI platforms now.
- Superblocks, AWS sign multi-year deal for enterprise AI on Bedrock
Superblocks, AWS sign multi-year deal for enterprise AI on Bedrock verdict.co.uk
- HUMAIN, MOZN to deliver enterprise AI solutions for financial sector
HUMAIN, MOZN to deliver enterprise AI solutions for financial sector verdict.co.uk
- Lead Data Engineer - Physical AI Platform, Data Engineering
Lead Data Engineer - Physical AI Platform, Data Engineering Built In
- AI cloud startup Volta valued at $2.4 billion, announces $10 billion AI partnership
AI cloud startup Volta valued at $2.4 billion, announces $10 billion AI partnership Reuters
- Eight-month-old UK start-up strikes $10bn deal with Anthropic
Eight-month-old UK start-up strikes $10bn deal with Anthropic The Telegraph
- Anthropic locks in $10 billion of compute from Volta, a cloud startup that didn't exist six months ago
Anthropic is locking in $10 billion worth of computing capacity from Volta Infra Holdings, a cloud startup that's only a few months old. The article Anthropic locks in $10 billion of compute from Volta, a cloud startup that didn't exist six months ago appeared first on The Decoder .
- The AI Boom’s Latest Winner: A Brand-New Startup That Just Landed a $10 Billion Deal With Anthropic
Volta was only established at the start of this year, and yet it’s suddenly a key player in the AI game.
- Glasp MCP Connector
Search your highlights and notes inside Claude and ChatGPT
- HappyRobot lands $150M Series C to scale agentic AI for enterprise operations
HappyRobot, a company developing agentic AI for supply chains has raised $150 million in Series C funding led by Prysm Capital and co-led by Eurazeo. Existing investors a16z, Base10, Y Combinator are ...
- HappyRobot is worth $1.2 billion. Its founder says it’s just ‘getting started’
HappyRobot is worth $1.2 billion. Its founder says it’s just ‘getting started’ Fortune
- HappyRobot Series C mints FreightTech’s newest unicorn
HappyRobot has raised about $200 million in 20 months, grown 5x since its Series B, and now runs AI agents inside more than 150 enterprise customers, from DHL to LKW WALTER. The post HappyRobot Series C mints FreightTech’s newest unicorn appeared first on FreightWaves .
- Agentic AI Startup Kily Raises ₹30 Cr To Accelerate Platform Deployment
AI startup Kily has raised a ₹30 Cr ($3.2 Mn) in a funding round led by early stage VC firm…
- Agentic AI startup Kily raises Rs 30 Cr in round led by Sorin Investments
Agentic enterprise AI startup Kily has raised Rs 30 crore ($3.1 million) in a funding round led by Sorin Investments, with participation from Razorpay and Wyser Capital. The fresh capital will be used to strengthen its product capabilities, expand go-to-market operations and increase adoption among consumer brands in India. Founded in 2025 by Sankalp Mehrotra, Anurag Singh and Sharad Chitlangia, Kily develops AI agents that help brands and sellers manage operations across e-commerce and quick commerce platforms. Its platform collects marketplace data and combines it with a brand's business goals and operating requirements to automate decisions and workflows. It helps brands manage areas such as product visibility, advertising, pricing, inventory and other commerce operations across digital platforms. Kily works across e-commerce and quick commerce platforms including Amazon, Flipkart, Blinkit, Zepto and Swiggy Instamart. Since its launch, the startup has worked with consumer brands including ITC. According to Kily, its AI agents are designed to reduce the use of spreadsheets, dashboards and manual processes by automating commerce management and decision-making.
- AI Startup Kily Bags INR 30 Cr in Sorin Investments-led Round
AI Startup Kily Bags INR 30 Cr in Sorin Investments-led Round india.entrepreneur.com
- AI-Native CRM Startup Superleap Bags ₹36 Cr Led By Peak XV’s Surge
Enterprise tech startup Superleap has raised ₹36 Cr (around $4.2 Mn) in its pre-Series A funding round led by Peak…
- Former Swiggy, Zomato executives launch AI startup Profound with $1.5 Mn seed round
Bengaluru based AI startup Profound has raised $1.5 million in a seed funding round as it launched an AI powered platform for professional networking and hiring. The startup was founded by Anuj Rathi and Prashant Parashar, former product and technology leaders with experience across Swiggy, Zomato, Cleartrip, Flipkart, Ola and Walmart Labs. The seed round saw participation from angel investors including Swiggy CEO Sriharsha Majety, Swiggy cofounder Nandan Reddy, former Zomato cofounder Pankaj Chaddah, Razorpay CEO Harshil Mathur, WhatsApp Global Head Kunal Shah, and OfBusiness cofounder Bhuvan Gupta, along with venture capital firms Stellaris Venture Partners and 3one4 Capital. The company will use the fresh capital to expand its engineering and product teams, enhance its AI capabilities, and develop its matching and introductions engine. It plans to onboard one million professionals globally before opening beta access to more users. Profound offers every professional an AI powered representative that starts with a voice conversation to understand a user's experience, expertise, work style and career goals. Based on this, the platform recommends opportunities and facilitates professional introductions. The startup also enables hiring managers to create AI representatives for roles and teams, allowing companies to capture hiring requirements through voice conversations instead of traditional job descriptions. "Just as actors have agents and athletes have managers, Profound gives each professional their own AI Rep," cofounder and CEO Anuj Rathi said in a statement. The company has opened early access for its first cohort, while a wider beta launch is planned later this year.
- Vertex Growth leads X Mile’s US$21.4M Series C for non-desk workforce AI
Japan’s logistics yards, construction sites and factory floors are not usually where venture capital headlines begin. Yet they sit at the centre of one of the country’s most urgent economic questions: how to keep essential industries running as the workforce ages, labour becomes harder to find, and smaller operators struggle to modernise. Tokyo-based X Mile […] The post Vertex Growth leads X Mile’s US$21.4M Series C for non-desk workforce AI appeared first on e27 .
- Zenity secures $125m Series C funding for AI agent governance
Zenity has completed a Series C round to support its ambitions to expand globally and further develop its AI security and governance platform.
- Zenity secures $125m Series C funding for AI agent governance
Zenity secures $125m Series C funding for AI agent governance verdict.co.uk
- Obsidian Security raises $85M as AI agents create cybersecurity’s next major attack surface
Obsidian Security Inc. has raised an $85 million Series D funding round at a post-money valuation of $1.1 billion as enterprises increasingly look to secure autonomous artificial intelligence agents accessing cloud applications, Chief Executive Hasan Imam said today in an exclusive interview with theCUBE with me at our NYSE Wired studio in NYC. The round […] The post Obsidian Security raises $85M as AI agents create cybersecurity’s next major attack surface appeared first on SiliconANGLE .
- Legal AI startup Aavalynx raises £1.5M to cut the cost of corporate disputes
Legaltech Aavalynx has raised £1.5 million in pre-seed funding to help enterprises tackle the financial risk of legal disputes. Founded in 2023 and commercially live in 2024, Aavalynx has developed p...
- Paris-based Shiplog raises over €807.6K to build agentic customer intelligence for personalisation at scale
Shiplog, a Paris-based startup building agentic customer intelligence for personalisation at scale, has raised over €807.6k ($930k) in pre-Seed funding to build what it believes will become the intelligence layer behind every customer interaction on the internet. The round was backed by Kima Ventures and Project Europe, alongside Purple, No Label Ventures, 100IN and Station […] The post Paris-based Shiplog raises over €807.6K to build agentic customer intelligence for personalisation at scale appeared first on EU-Startups .
- Alibaba launches next-gen Qwen3.8 with 2.4 trillion parameters
Alibaba on Monday unveiled its next-generation foundation model Qwen3.8. The model features 2.4 trillion parameters and delivers improvements in coding and professional workplace tasks (Cowork). In the recently released third-party Arena rankings, Alibaba’s Qwen model ranked second only to Anthropic’s Claude series. The Qwen3.8 API is now available through the Qwen AI platform and has […]
- Alibaba Releases New Qwen Model, Consolidates AI Office Tools
Alibaba Releases New Qwen Model, Consolidates AI Office Tools Caixin Global
- Tech Brief (Aug. 4): Alibaba Launches New Large Model to Power Enterprise-Facing AI Agent
Tech Brief (Aug. 4): Alibaba Launches New Large Model to Power Enterprise-Facing AI Agent Caixin Global
- Alibaba Unveils Biggest Qwen Model; DeepSeek Drives AI Costs Lower
Alibaba Unveils Biggest Qwen Model; DeepSeek Drives AI Costs Lower apac.entrepreneur.com
- Computing Actual Causes for Neural Network Predictions under Structured Causal Inputs
Explaining the predictions of neural networks is a central challenge in trustworthy AI. Existing explanation methods, such as those based on feature attribution or minimal sufficient sets, typically treat input features as independent, which can yield misleading explanations when inputs exhibit stru...
- Can LLMs Test Terminal User Interfaces?
Terminal User Interfaces (TUIs) combine the stateful, screen-oriented behaviour of GUIs with terminal deployment and are now common in developer tools. Yet they lack a dedicated testing methodology. We survey 197 real-world TUI applications: only 12% of test code exercises the interface, and 45% of ...
- AI-Based Sound Effect Generation: A Narrative Review of Generative Models Across Input Modalities
Sound effects play a crucial role in conveying actions, events, and environmental cues across digital applications, often requiring a high degree of variation and contextual adaptability. Artificial intelligence (AI)-driven audio generative models are rapidly growing in popularity and have the poten...
- Failure-Informed Image Self-Augmentation for Multimodal Large Language Model Self-Improvement
Multimodal large language models (MLLMs) have achieved remarkable performance across vision-language tasks, but their progress depends heavily on large-scale, high-quality multimodal data that are costly to annotate. Self-augmentation offers a promising alternative by enabling models to expand their...
- CARE-Bench: Benchmarking Patient-Facing LLM Triage
Patient-facing medical LLMs and agents increasingly answer symptom questions before clinician contact, where the key safety question is what action the user should take next. We introduce CARE-Bench, a source-grounded benchmark that evaluates sequential patient-facing triage as a four-label per-turn...
- GPTKB 2.0: Direct Construction of Disambiguated Knowledge Bases from Large Language Models
Automated Knowledge Base Construction (AKBC) is a core NLP task, and recent work proposes generating knowledge bases directly from large language models (LLMs), treating the model itself as the knowledge source. However, LLMs natively possess no representation of entities, leading to duplicate entri...
- When Outputs Disperse, Does Epistemic Revision Follow? A Black-Box Coupling Diagnostic for Machine Collectives
Collective intelligence research treats disagreement as evidence of epistemic diversity: if agents express different views, the group should retain capacity to revise. In LLM collectives this proxy can break: agents can produce diverse-looking arguments while preserving the same conclusion. We opera...
- Pattern over Pixels: Measuring Pattern Completion Bias in Multimodal Code Generation
Multimodal large language models (MLLMs) are increasingly used to translate webpage screenshots into front-end code, but repeated UI patterns may sway them toward visually incorrect yet pattern-consistent outputs. In this work, we test how repeated webpage patterns hurt MLLM accuracy on an objective...
- PhyAI: Real-Time Physical AI at the Edge, Scalable Rollouts in the Cloud
Physical AI policies require inference throughout their lifecycle, including model evaluation, cloud reinforcement learning rollout, edge GPU serving, and onboard deployment. Although these settings share the same checkpoint and action semantics, they often rely on separate inference programs. To un...
- space ocr
OCR that checks its own answers, as an app or an API
- Taming the Implicit: Dual-Channel Risk-Aware Reinforcement Fine-Tuning for Continual Multimodal Post-Training
Reinforcement fine-tuning (RFT) is widely believed to inherently resist catastrophic forgetting in continual post-training of multimodal large language models. Under pronounced task distributional shifts, however, forgetting across representative RFT algorithms escalates sharply. This stems from the...