COGNOSCERE Daily Tech Review — Issue T189 · Sunday, July 26, 2026

Or visit Intelligence Overview for deeper analysis.

ACTCertighost (CVE-2026-54121): Public Exploit Released for Active Directory Domain Takeover Flaw

Security researchers published a working proof-of-concept exploit for CVE-2026-54121 (Certighost) on July 24, demonstrating that any low-privileged Active Directory domain user can impersonate a Domain Controller via an AD CS enrollment fallback flaw, perform a DCSync attack, and extract the krbtgt secret to fully compromise a Windows domain. Microsoft patched the issue in its July 14, 2026 updates; organizations that have not applied the July patches are immediately at risk as no in-the-wild exploitation has been confirmed yet but a functional PoC is publicly available.

The Hacker News · Cybersecurity · Relevance: 0.9 · Source →

Active Directory, AD CS, privilege escalation, domain compromise, Windows Server, patching, zero-day, PoC exploit

ACTOracle Critical Patch Update, July 2026: Record 1,449 Security Patches

Oracle released the largest Critical Patch Update in its history, shipping 1,449 security patches covering 1,235 unique CVEs across 32 product families including E-Business Suite, Fusion Middleware, and Database Server; 261 patches carry critical severity ratings and approximately 600 flaws are remotely exploitable without authentication. Oracle Fusion Middleware alone received 355 patches, ten of which scored a perfect 10.0 CVSS.

Security Boulevard · Cybersecurity · Relevance: 0.9 · Source →

patching, vulnerability management, enterprise software, ERP, critical patch, remote code execution

PREPAREClaude Cowork Flaw Could Let AI Agent Escape Its VM and Access Mac Files

Researchers at Accomplish AI disclosed a sandbox escape vulnerability (SharedRoot) in Anthropic's Claude Cowork macOS application: by exploiting Linux kernel flaw CVE-2026-46331, an AI agent running inside the local Linux VM can gain guest-root and read or write files across the entire host Mac, including SSH keys and cloud credentials, affecting an estimated 500,000 local users. Anthropic did not issue a direct patch but mitigated the issue by defaulting Cowork to cloud execution; users running local sessions remain exposed unless additional kernel hardening is applied.

The Hacker News · Cybersecurity · Relevance: 0.9 · Source →

AI agent security, sandbox escape, VM isolation, macOS, credential exposure, agentic AI risk, enterprise security

WATCHPsiQuantum Signs $125 Million Agreement with DARPA Under Quantum Benchmarking Initiative

Photonic quantum computing startup PsiQuantum signed a $125 million expanded agreement with DARPA on July 22, 2026 under the Quantum Benchmarking Initiative's final Stage C, the company's largest US government award to date, to fund independent verification of its silicon-photonic hardware architecture, software, and infrastructure. PsiQuantum is one of only two companies in the program's final stage alongside Microsoft, with DARPA teams now embedded inside the company; the company's CEO anticipates a commercial-scale machine operational around 2030.

Reuters / Yahoo Finance · Emerging Technology / Quantum Computing · Relevance: 0.8 · Source →

quantum computing, photonic computing, government funding, DARPA, emerging technology, cryptography risk, defense

PREPAREAMD and Anthropic Announce Strategic Partnership to Deploy Up to 2 Gigawatts of AMD Instinct MI450 Series GPUs

AMD and Anthropic announced a strategic partnership on July 22, 2026 under which Anthropic will deploy up to 2 gigawatts of AMD Instinct MI450 Series GPUs in AMD Helios rack-scale systems beginning in the first half of 2027, with AMD committing a milestone-contingent equity investment of up to $5 billion in Anthropic. The two companies will also launch a multi-year engineering collaboration to use Claude to optimize AMD GPU workloads and accelerate ROCm software development, positioning AMD more competitively against Nvidia's CUDA ecosystem.

AMD Newsroom · AI Infrastructure / Enterprise AI · Relevance: 0.9 · Source →

AI chips, GPU infrastructure, AI compute, Nvidia alternative, AI training, strategic investment, data center

PREPARESierra Acquires Takeoff, Launches Horizon Platform for Long-Horizon AI Agents

Sierra, the enterprise AI agent company led by former Salesforce CTO Bret Taylor and valued at $15 billion, acquired Takeoff on July 23, 2026, a 14-month-old startup that grew from zero to nearly eight-figure ARR with three employees by building long-horizon agent runtimes capable of executing multi-step tasks across days or weeks in sectors including lending, healthcare, and telecom. The two companies are building a combined platform called Horizon, which can orchestrate outbound and inbound interactions over extended timeframes under an outcomes-based pricing model.

CMSWire · Agentic Systems / Enterprise AI · Relevance: 0.9 · Source →

AI agents, enterprise automation, long-horizon agents, M&A, agentic AI, outcomes-based pricing, workflow automation

WATCHStripe in Talks to Acquire AI Model Marketplace OpenRouter for ~$10 Billion

Stripe is in preliminary discussions to acquire OpenRouter, an AI model-routing marketplace that lets developers access over 400 large language models from 70-plus providers through a single API, at a valuation of approximately $10 billion — nearly eight times OpenRouter's $1.3 billion May 2026 valuation. The deal, reported by The Wall Street Journal, would mark Stripe's most significant expansion beyond payments into AI infrastructure, though talks remain ongoing and could still collapse.

The Next Web · AI Tooling / M&A · Relevance: 0.9 · Source →

M&A, AI infrastructure, model routing, fintech, AI marketplace, LLM access, developer tools

WATCHBlack Forest Labs Launches FLUX 3: Unified Multimodal Model for Video, Audio, Image, and Robot Action Prediction

Black Forest Labs launched FLUX 3 on July 23, 2026, a single multimodal foundation model jointly trained on images, video, and audio that generates 20-second video clips with native synchronized audio, and extends the same architecture to robot action prediction via FLUX-mimic, which is already deployed on Audi production lines after learning factory tasks from roughly 30 minutes of demonstration data. FLUX 3 Video and Action are in gated early access; the image module and an open-weight Dev release are planned for later in 2026.

VentureBeat · AI Model Releases / Robotics · Relevance: 0.8 · Source →

multimodal AI, video generation, generative AI, robotics, physical AI, content creation, open-weight models

ACTEU AI Act Transparency Obligations Take Effect August 2, 2026: Enterprises Deploying AI in EU Must Comply

The EU AI Act's transparency obligations for AI systems that interact with users become enforceable on August 2, 2026, requiring enterprises to notify EU-based users when they are interacting with AI and imposing penalties of up to €35 million or 7% of global annual turnover for breaches of prohibited practices already in force. A simplification package has pushed standalone high-risk AI system obligations to December 2027, but the upcoming August deadline affects any company deploying consumer-facing AI in Europe regardless of where the company is headquartered.

vishleshan.ai · Regulatory Risk / AI Governance · Relevance: 0.9 · Source →

EU AI Act, AI regulation, compliance, transparency, AI governance, risk management, enterprise AI policy

PREPAREChina AI Agent Rules and Illinois Frontier AI Audit Law Create Dual Enterprise Compliance Deadlines

China's Implementation Opinions on intelligent agents became enforceable July 15, 2026, establishing the world's first dedicated regulatory category for AI agents with a three-tier decision authorization framework and mandatory filing requirements for high-risk sectors. Separately, Illinois enacted the first US state law requiring annual independent safety plan audits for frontier AI model developers with over $500 million in revenue, creating structural governance requirements that enterprise compliance and legal teams cannot treat as future-state concerns.

AI Governance Institute · Regulatory Risk / AI Governance · Relevance: 0.8 · Source →

AI regulation, AI agents, China tech policy, US state law, compliance, AI governance, frontier AI, audit requirements