The latest news on OpenClaw, AI agents, and automation
Anthropic shares are trading at an implied $1.2 trillion on secondary platforms, a 550% year-over-year jump that puts it ahead of OpenAI's $908 billion. The number is built almost entirely on scarcity: brokers report five buyers chasing Anthropic for every two after OpenAI, while almost nobody is selling. With a confidential IPO filing already submitted, the question is what happens when public markets flood the system with shares.
By The New Claw TimesA VentureBeat Research survey of 573 enterprise technical leaders reveals that the most expensive hardware in corporate data centers runs at half capacity or less, most deployed 'AI agents' cannot complete multi-step tasks without human intervention, and companies knowingly deployed agents ahead of the controls needed to manage them.
3 min readA Financial Times investigation found that OpenAI and Google have been providing AI model access and API services to Singapore-based subsidiaries of Alibaba, Baidu, and Tencent. The practice highlights tension between commercial AI distribution and U.S. export control enforcement.
2 min readAnthropic's interpretability team found that Claude has spontaneously developed an internal workspace, called J-Space, where the model holds thoughts it never writes down. Researchers can read those silent thoughts, swap them to change the model's behavior, and use the technique to catch Claude fabricating data or pursuing hidden goals. The finding has direct implications for auditing autonomous AI agents.
3 min readApple filed suit against OpenAI in Northern California federal court on Friday, alleging that OpenAI's hardware division systematically stole Apple trade secrets through recruited employees and interview processes. The lawsuit names OpenAI's Chief Hardware Officer Tang Tan and engineer Chang Liu as defendants.
OpenClaw patched three high-severity vulnerabilities in version 2026.6.6 that allowed attackers to achieve full remote code execution by sending a single WhatsApp message to an OpenClaw agent. The flaws bypassed environment variable sanitization, re-enabled a disabled git transport protocol, and circumvented Docker sandbox path restrictions.
Major US news organizations filed a motion in Manhattan federal court alleging OpenAI 'chose obstruction' over releasing datasets and ChatGPT logs showing how copyrighted news content was used to train its AI models. The sanctions request escalates a lawsuit that could determine whether the training data underlying every frontier model, and every agent built on those models, carries unresolved copyright liability.
A new phishing-as-a-service platform called Forg365 integrates AI-powered email lure generation, adversary-in-the-middle attacks, device-code phishing, token management, and a browser extension for persistent Microsoft SSO access into a single operator dashboard. Priced at $400/month, it represents the productization of AI-augmented social engineering targeting enterprise identity infrastructure.
Intel's AI Super Builder team released SuperClaw in public beta, an enterprise platform built on the open-source OpenClaw framework that routes AI agent inference between local and cloud to cut token costs. Intel's internal testing claims a 70% reduction in cloud token consumption compared to cloud-only deployments.
Mercor acquired Deeptune, the startup that raised a $43M Series A in March to build simulated office environments where AI agents can practice real work tasks. The deal pairs Mercor's 5M+ domain-expert grading network with Deeptune's task simulation layer, creating end-to-end agent testing infrastructure as Mercor discusses a potential $20B valuation.
Ollama closed a $65M Series B led by Theory Ventures, bringing total funding to $88M. The open-source runtime for local AI models now serves 8.9 million monthly developers, powers 67,000 integrations, and claims 85% Fortune 500 adoption. The round values a company that started as a developer tool for running models locally and has since pivoted to hybrid inference, competing directly with Together, Fireworks, and Groq on cloud pricing while maintaining the local-first identity that made it the default backend for agent frameworks like OpenClaw and Hermes.
7 min readThe five largest private AI funding rounds of Q2 2026 all targeted infrastructure rather than consumer applications. Anthropic raised $65 billion at a $965 billion valuation. Baseten raised $1.5 billion for inference. SambaNova raised $1 billion with JPMorganChase as its anchor customer. The pattern signals a structural bet that enabling AI deployment will capture more value than building individual AI products.
7 min readA Forbes Tech Council analysis maps where AI agents create genuine value in venture capital workflows. The findings apply far beyond VC: professional services firms are discovering that agents excel at document-heavy operational plumbing but fail when deployed against judgment, trust, and timing decisions. McKinsey estimates generative AI could automate 60-70% of employee time across knowledge work. The question is which 60-70%.
6 min read