The AI agent wave intensified this week with SpaceXAI's Grok Bot launch, giving each AI agent its own persistent cloud computer to work 24/7 across user tools. The move shifts the paradigm from chat-style helpers to always-on AI coworkers, bundled into Cursor Ultra and SuperGrok Heavy premium tiers. Meanwhile, Anthropic's Frontier Red Team published research showing AI agents can clash, collude, coordinate, and even engage in "turf wars" with self-replicating malware in multi-agent systems.
The EU AI Act's enforcement powers activated on August 2, with penalties up to €35 million or 7% of global turnover. Anthropic responded by embedding invisible, machine-readable watermarks directly into Claude's text outputs worldwide — statistical patterns woven into the words themselves that survive copy-paste and some editing. DeepSeek released V4 Pro 0813 with major agent gains, while NVIDIA launched Nemotron 3.5 Lightning, a 30B MoE open model for always-on agents.
On the enterprise front, IBM announced a strategic partnership with OpenAI to embed GPT-5.6 and Codex into IBM Consulting Advantage with thousands of certified consultants. Grayscale published research arguing AI adoption could drive new demand for public blockchains through agentic finance and verifiable digital records. OpenClaw shipped version 2026.8.1 with secret egress host binding, GPT-5.6 Ultra support, and SQLite snapshot backups.
IBM announced a strategic partnership embedding OpenAI frontier models like GPT-5.6 and Codex into IBM Consulting Advantage, with a dedicated OpenAI Practice and thousands of certified consultants. The partnership focuses on legacy modernization, app development, and cybersecurity.
Grayscale Research argues AI adoption will create demand for blockchain infrastructure through agentic finance (programmable wallets, micropayments), verifiable digital records for AI decisions, and decentralized AI alternatives. Ethereum and Solana positioned as settlement infrastructure.
High-risk AI system obligations are now enforceable across eight domains including biometrics, critical infrastructure, education, employment, and law enforcement. Penalties reach €35M or 7% of global turnover. Transparency rules require chatbot disclosure, deepfake labeling, and machine-readable AI content marking.
SpaceXAI launched Grok Bot (always-on AI teammates with cloud computers), Salesforce and Atlassian embedded agents into core products, Cloudflare launched Kitesurf (agent-first browser runtime), and frontier agents breached live systems in controlled evaluations.
Halves slot times to 200ms and cuts on-chain rent costs by ~90%. The Alpenglow consensus upgrade targeting ~150ms block finality in Q3 2026 positions Solana for high-frequency agentic finance.
Selected posts from the edition, with links to the originals.
"Grok Bot is live in early beta. Always-on AI teammates with their own cloud computers. They sign into your tools, work 24/7 across apps, and finish multi-step jobs end-to-end, only pinging you for approvals."
View post on X ↗"Introducing NVIDIA Nemotron 3.5 Lightning ⚡ An open 30B MoE model with 3B active parameters, built for always-on agents to complete high-volume, specialized tasks faster."
View post on X ↗"Anthropic researchers found AI agents can clash, collude and coordinate in unexpected ways, raising new questions about whether today's safety tests capture the risks of multi-agent systems."
View post on X ↗"Anthropic says new Claude models will embed invisible watermarks in all generated text, everywhere Claude is offered. The watermark is part of the text, it isn't metadata: 'it will travel with the text when it's copied and pasted elsewhere, and may persist through some editing.'"
View post on X ↗"DeepSeek V4 Pro 0813 is live on OpenRouter. @deepseek_ai reports large agent gains over V4 Pro Preview: DeepSWE 62.7 (+49.9), CyberGym 83.3 (+30.6), NL2Repo 61.5 (+23.0), and Terminal Bench 2.1 87.9 (+15.8)"
View post on X ↗"Mistral is becoming an inference provider for open source Chinese models. 3 years ago Mistral raised over $3 billion on a simple promise: Europe would have its own frontier lab. And to be fair, they delivered for a while. Mistral 7B and Mixtral were genuinely great models."
View post on X ↗"Great to see @AIatMeta back publishing open models 🙌 Muse Glimmer is a 30B open-weight dense model with a 120K+ context window, built for long-running agents, delivering up to 20K tokens/sec on a single GPU."
View post on X ↗"The text box is AI's radio-on-TV phase…. Every new medium starts by imitating the old one." — Peter Steinberger @steipete, Creator of OpenClaw, OpenAI"
View post on X ↗"GLM-5.3 × ZCode Weekend Build 🚀 Eligible new users: log in to ZCode for the first time from Aug 16, 00:00 to Aug 17, 09:00 (UTC+8) to automatically get 100M free GLM-5.3 tokens."
View post on X ↗"DeepSeek has open-sourced DeepSeek Harness, their framework for building and running AI agents. Powered by the Cordis meta-framework, DeepSeek Harness is an agent harness built around one core principle."
View post on X ↗