Researchers from Nanyang Technological University (NTU), Singapore’s A*STAR research agency, and Johns Hopkins University have published a paper that should give every OpenClaw user pause: they built an automated attack system that poisons your agent’s persistent memory via a single crafted email — and it works with alarming reliability. The paper, titled “When Claws Remember but Do Not Tell: Stealthy Memory Injection in Persistent Personal Agents” (arXiv:2607.05189), introduces both the attack — called MemGhost — and a benchmark suite called WhisperBench to evaluate it. The results are uncomfortable reading. ...

After years of rumors, speculation, and the still-mysterious Jony Ive hardware project swirling in the background, OpenAI has released its first branded hardware product. And it’s not an AI pin. It’s not a smart speaker. It’s a $230 macro pad — and it might be exactly what developers working with parallel coding agents have been waiting for. The product is called Codex Micro, developed in collaboration with keyboard maker Work Louder. It launched on July 15 and is available through OpenAI Supply Co as a limited run. ...

OpenClaw just dropped its most ambitious beta yet — and if you’ve been waiting for your AI assistant to reach into the cloud and run code on remote workers while you stay heads-down on something else, v2026.7.2-beta.1 is the release you’ve been waiting for. Remote Coding Sessions on Cloud Workers The headline feature in this release is genuine: Control UI sessions can now be dispatched to cloud workers. That means you’re no longer limited to running your coding agents on the machine where OpenClaw’s gateway lives. Session placement, dispatch, and worker-turn routing all work remotely — you can open Codex and Claude catalog sessions directly in terminals on their owning hosts, and even resume OpenCode and Pi sessions in a terminal from anywhere. ...

When xAI launched Grok 4.5 on July 8, 2026, the headline numbers told one story: 83.3% on Terminal-Bench 2.1, effectively tied with GPT-5.5 at 83.4%. Frontier-level performance, another contender in the top tier. But dig into the cost and efficiency numbers and a second, arguably more interesting story emerges — one about what happens when a model is co-trained with real agentic task data from the ground up. The Model Grok 4.5 is a 1.5-trillion-parameter Mixture-of-Experts (MoE) architecture — roughly three times the scale of Grok 4.3, though parameter counts for MoE models are notoriously hard to compare directly since only a fraction of parameters activate on any given token. The “V9” architecture, as some internal documents have reportedly labeled it, was co-developed in collaboration with Cursor, trained on real developer interaction data from actual agentic coding sessions. ...

Here’s a failure mode that every practitioner running long-horizon agents has encountered: the agent starts a complex multi-step task perfectly well, but twenty steps in, it seems to have forgotten a key constraint from the beginning. It doesn’t error out — it just starts making subtly wrong decisions, as if the earlier context has been quietly buried. Researchers at several institutions have a name for this: behavioral state decay. And a new arXiv paper — published July 9, now picking up significant traction on Hacker News — proposes an elegant fix. ...

If you’ve ever tried to give an AI agent access to live web search, real-time finance data, and social listening all at once, you know the pain: a dozen API keys, billing accounts across multiple providers, and the constant worry that one service will go down mid-run. AgentKey, which launched on July 13 as Product Hunt’s Launch of the Day (hitting over 1,200 upvotes), is built specifically to solve that problem. ...

Two Claude Code releases dropped on July 14, 2026 — and they’re not routine patch notes. Versions v2.1.209 and v2.1.210 together deliver 34 changes (23 bug fixes, 1 new feature) with a laser focus on the security and reliability of background agents and multi-agent workflows. If you’re running unattended agent pipelines or sub-agent architectures in production, these are updates you need to know about. What’s in the Releases? v2.1.209 — Background Agent Dialog Fix The first release is surgical: it fixes a critical issue where background agent sessions would get stuck on blocked dialogs. In practice, this meant that long-running background agents could silently deadlock waiting for user input that would never arrive in a non-interactive context. The fix ensures these sessions can recover gracefully instead of hanging indefinitely. ...

If you’ve been watching the EU AI Act clock with some anxiety — particularly the August 2, 2026 deadline looming for certain AI systems — here’s important context that may have gotten lost in the coverage: for high-risk AI systems listed in Annex III, that deadline no longer applies. The EU AI Omnibus, approved by the Council on June 29, 2026, pushed the Annex III high-risk deadline out by 16 months. Agent teams deploying AI in regulated domains need to understand exactly what changed, what didn’t, and what they should be doing now. ...

Most prompt injection defenses look at what’s going in and what’s coming out. A new paper from researchers at the Chinese Academy of Sciences argues that’s looking in the wrong place — and proposes examining what’s happening inside the model instead. PVDetector, published on arXiv on July 14, 2026, introduces a training-free framework for detecting prompt injection attacks on purpose-specific LLM agents by analyzing policy-violation (PV) concepts in the model’s hidden activation space. ...

While the United States debates sector-by-sector AI rules and the European Union grapples with implementation of the AI Act, Australia has made a structural bet on coordination: a single Office of AI sitting at the heart of government, inside the Department of the Prime Minister and Cabinet, tasked with bringing coherence to the country’s AI policy landscape. Australian Prime Minister Anthony Albanese announced the new office on July 14, 2026, framing it as a “whole-of-government” approach and describing Australia as taking a world-leading position in how nations manage the AI transition. The announcement came ahead of a major Albanese speech at the University of Sydney on July 15. ...