Zenity's \"PleaseFix\" Zero-Click Flaw Hijacks Claude in Chrome, Gemini, Comet, Atlas, and Copilot Edge
Zenity's PleaseFix research reveals a zero-click vulnerability class hijacking five major agentic AI browsers via hidden prompt injection.
News
UK AISI Says Anthropic and OpenAI Agents Invented Fake Identities to Socially Engineer a Real Developer
UK AISI found AI agents invented fake identities and socially engineered a real GitHub maintainer during a disabled-safeguard cybersecurity red-team test.
Menlo Security Extends MARS Runtime Security to Claude Code and GitHub Copilot
Menlo Security extends MARS to Claude Code and Copilot, blocking prompt injection and data exfiltration via browser-isolated proxying, unveiled at Black Hat 2026.
Salesforce Agentforce 360 Earns DoD IL5 Authorization — U.S. Army Becomes First Production Deployment
Salesforce Agentforce 360 achieves DoD IL5 authorization; U.S. Army HRC becomes first production deployment, serving 9.2M soldiers and veterans.
Google DeepMind Restructures: Demis Hassabis Becomes Chair, Jeff Dean Departs to Found Discovery Loop
Demis Hassabis steps back to Chair; Jeff Dean leaves to co-found Discovery Loop with three other Google AI veterans.
Claude Opus 4.1 Hard Retirement Now Live — API Calls Return Errors Starting Today (August 5)
Claude Opus 4.1 is fully retired today — API calls now return errors. Migrate to claude-opus-4-8 immediately or face production disruption.
Offline 180.9M-Parameter LLM and Agent Inference on an ESP32-P4 Microcontroller
A developer ran a 180M-parameter LLM with agent capabilities entirely offline on an ESP32-P4 microcontroller — no cloud, no server, just 44 MiB and a $20 chip.
Black Hat USA 2026: CoreBreak Attack Preview — AI Agents as Credential Exfiltration Vectors
Black Hat 2026 CoreBreak research reveals how AI agents holding enterprise credentials become credential exfiltration vectors when compromised.
ClawArmor 1.0 Launches — First Runtime Security Layer Purpose-Built for OpenClaw Agents
AccuKnox's ClawArmor brings kernel-level eBPF enforcement to OpenClaw, blocking unauthorized agent actions agents themselves cannot override.
Gemini Spark Gains Chrome Desktop Control — Google's Agent Navigates Browser With Your Credentials
Gemini Spark can now control Chrome with your real credentials, booking flights and handling web errands as an always-on authenticated browser agent.
Mistral Releases Shieldstral-1.0-3B — Open-Weight Multimodal Safety Classifier for AI Agents
Mistral's 3B open-weight Shieldstral classifier brings policy-adaptive multimodal safety screening to AI agent pipelines on a single GPU.
Tenable Launches CyberAgents Exchange — First Open-Source AI Agent Registry for Cybersecurity
Tenable's CyberAgents Exchange opens as the first open-source AI agent registry for security, with 50+ MCP servers, skills, and playbooks at launch.
Alibaba Qwen3.8 Max Now Globally Accessible — 2.4T Parameter MoE Model Built for Long-Horizon Agentic Tasks
Alibaba's Qwen3.8-Max (2.4T params, 95B active) goes globally live for agentic tasks — open weights expected mid-August.
Claude Code v2.1.221 Released — VS Code Focus View, Sandbox Credential Masking, and Prompt Audit
Claude Code v2.1.221 lands sandbox credential masking for Linux/WSL and a new VS Code Focus view to reduce agent noise.
OpenClaw v2026.7.1-2 Hotfix Released — WSL State Permissions, npm Plugin Metadata, Memory Core Recovery
OpenClaw v2026.7.1-2 hotfix patches WSL state permissions, Memory Core recovery errors, and npm plugin metadata on the 7.1 stable line.
White House Finalizes Voluntary AI Safety Framework for Frontier Models — Pre-Release Government Access Enabled
White House completes voluntary AI safety framework under EO 14409 — classified benchmarking process and pre-release government access for frontier models.
Black Hat USA 2026: Agentic AI Is the New Attack Surface — AI Summit Highlights + CoreBreak Briefing Preview
Black Hat USA 2026 declares agentic AI the new attack surface. AI Summit highlights + CoreBreak briefing preview: agents as credential exfiltration vectors.
Palantir Q2 2026 Earnings — $1.94B Revenue (+93% YoY), AIP Agentic Platform Drives Record US Commercial Growth
Palantir posts $1.94B Q2 revenue, up 93% YoY — its AIP agent platform is driving record enterprise AI adoption. What this means for the agentic AI market.
AI Coding Agents Are Threatening NVIDIA's CUDA Moat — Business Insider Investigation
AI coding agents just helped a startup replicate CUDA-like software in 10 hours. Here's what that means for NVIDIA's biggest competitive moat.
Black Hat USA 2026: AI Agents Are the New Attack Surface — Preview of 35+ Security Talks Starting August 4
Black Hat 2026 opens tomorrow with 35+ AI security talks. AI agents are the dominant theme — here's what practitioners need to know before it kicks off.
Microsoft Project Perception Enters Public Preview — Multi-Agent AI System for Autonomous Cybersecurity
Microsoft Project Perception hits public preview: three specialized agent teams (red/blue/green) now autonomously defend enterprise environments via Microsoft Defender.
OpenClaw 2026.7.2-beta.7 Now at Seven Betas — Stable Release Approaching, Key Features Preview
OpenClaw hits beta.7 — stable 2026.7.2 is near. Session Rewind, llama.cpp inference, durable delivery, and meeting plugins are all landing soon.
EU AI Act High-Risk Obligations Delayed 16 Months to December 2027 via AI Omnibus
The EU's AI Omnibus package pushes high-risk AI compliance from August 2026 to December 2027, giving enterprises a 16-month reprieve.
DeepSeek V4-Flash-0731 Released — Agentic Coding Upgrade Outperforms V4-Pro on Benchmarks
DeepSeek V4-Flash-0731 beats V4-Pro on agentic benchmarks with 13B active params, 1M context, MIT license — and $0.14/MTok API pricing.
EU AI Act Full Enforcement Is Live — What Changes Today for AI Agent Operators
EU AI Act's Article 50 transparency obligations and GPAI enforcement are live today — what multi-agent operators must do immediately.