COGNOSCERE Daily Tech Review — Issue T227 · Thursday, September 3, 2026

Or visit Intelligence Overview for deeper analysis.

ACTAnthropic Warns Claude Users of Infostealer Malware Infections

Anthropic notified affected Claude users that infostealer malware variants — including Vidar, Lumma, StealC, RedLine, and AMOS — hijacked browser sessions and drained API usage limits; the company logged out compromised sessions, removed stored payment methods, and refunded unauthorized charges. Enterprises deploying Claude should require malware removal before re-enrolling payment credentials and review endpoint security posture.

TLDR InfoSec · Cybersecurity Incidents · Relevance: 0.8 · Source →

infostealer, malware, credential-theft, AI-platform-security, endpoint-security, browser-hijacking

PREPAREIntroducing Azure Multicloud Interconnect for AWS

Microsoft and AWS launched a joint multicloud networking solution combining Azure Multicloud Interconnect and AWS Interconnect — multicloud, enabling enterprises to provision private, high-performance connectivity between the two clouds via a streamlined, API-driven experience built on open specifications. The service enters public preview in four regions, with up to 100 Gbps throughput at general availability.

Microsoft Azure · Cloud Computing / Multicloud · Relevance: 0.9 · Source →

multicloud, cloud-networking, enterprise-infrastructure, Azure, AWS, hybrid-cloud, cloud-interoperability

PREPAREOpenAI to Cut Off Cursor's Access to Its Coding Models in November

OpenAI announced it will terminate its relationship with Cursor on November 12, cutting off access to its coding models roughly 10 weeks after SpaceX completed its $60 billion acquisition of Cursor's parent company — with OpenAI citing prior contract violations by Elon Musk's companies as the stated reason. Enterprise leaders are being warned that AI model availability inside SaaS tools can change suddenly due to acquisitions or competitive dynamics, making multi-model support increasingly critical.

CIO · Enterprise Software / AI Tooling · Relevance: 0.8 · Source →

AI-tooling, vendor-dependency, coding-agents, enterprise-software, model-access, SaaS-risk, developer-tools

WATCHPalo Alto Networks Acquires Console to Agentify Security

Palo Alto Networks acquired Console, an AI-native ITSM and automation platform, to expand agentic AI capabilities inside its Cortex security operations platform. Console connects to identity systems, SaaS applications, and security tools, enabling AI agents to autonomously gather context, execute workflows, and document remediation actions at machine speed.

Palo Alto Networks · Cybersecurity M&A · Relevance: 0.9 · Source →

cybersecurity-M&A, agentic-AI, ITSM, security-automation, SOC, enterprise-security, Cortex

PREPAREAnthropic's Claude Fable 5.1 and Mythos 5.1 Arrive with a 75% Cost Reduction for Fable Cache Reads

Anthropic released Claude Fable 5.1 and Mythos 5.1, its most capable models yet, with major gains on autonomous coding and research benchmarks, a 75% cut to cache-read pricing, and fewer safety rejections on cybersecurity and biology queries. Mythos 5.1 is a restricted-access variant for vetted U.S. cybersecurity and life-sciences organizations only.

Anthropic · AI Model Releases · Relevance: 1.0 · Source →

LLM, frontier-AI, enterprise-AI, pricing, agentic, coding-agent, AI-safety, model-release

PREPARECommission Designates ChatGPT, Reddit, Roblox Under Digital Services Act

The European Commission designated ChatGPT as a Very Large Online Search Engine (VLOSE) and Reddit and Roblox as Very Large Online Platforms (VLOPs) under the Digital Services Act, based on each service exceeding 45 million EU monthly users. The three services have until late November/December 2026 to comply with obligations including systemic risk assessments, algorithmic transparency, and independent audits, with non-compliance risking fines of up to 6% of global annual turnover.

European Commission · AI Regulation · Relevance: 0.9 · Source →

EU-regulation, DSA, AI-governance, compliance, platform-regulation, transparency, risk-assessment

WATCHOpenAI Says Astra AI Model Is Its First That Crosses 'Critical' Cybersecurity Capability

OpenAI announced its forthcoming Astra model is the first it has designated as meeting its 'Critical' cybersecurity threshold under its Preparedness Framework, meaning the model can autonomously find and exploit previously unknown security flaws without human guidance. Access to Astra's most advanced cyber capabilities will be restricted to a vetted group through its Daybreak Blue program.

CNBC · AI Safety & Governance · Relevance: 0.9 · Source →

AI-safety, frontier-AI, cybersecurity, model-release, autonomous-exploitation, preparedness-framework

WATCHAnthropic Signs $35 Billion Cloud Deal with Nvidia-Backed Lambda

Anthropic signed a $35 billion, six-year cloud computing agreement with Lambda — an AI cloud provider backed by Nvidia — for capacity at a Hut 8 data center in Nueces County, Texas, with Nvidia holding the facility lease and supplying chips. The deal is the latest in a series of massive infrastructure commitments by Anthropic, which has now locked in over $125 billion in compute capacity across multiple providers as demand for Claude products surges.

Bloomberg · AI Infrastructure · Relevance: 0.9 · Source →

AI-infrastructure, cloud-computing, data-center, GPU, compute-capacity, AI-supply-chain, Nvidia

WATCHAI Startup Cognition Set to Raise Around $1 Billion at a $47 Billion Value

Cognition AI — maker of the autonomous engineering agent Devin — is closing a new funding round of approximately $1 billion at a ~$47 billion valuation, nearly doubling its May 2026 mark of $26 billion, as annualized revenue has surpassed $900 million and investor demand for the round reached nearly $10 billion. The raise follows SpaceX's $60 billion acquisition of rival Cursor.

Bloomberg · AI Funding & Market · Relevance: 0.9 · Source →

AI-funding, agentic-coding, venture-capital, AI-agents, software-development, enterprise-AI, valuation

WATCHRunway Introduces Solaris: An Interface World Model

Runway launched Solaris in early access, an 'Interface World Model' that renders interactive websites and apps as real-time video — generating every frame and user-interaction response without any underlying code. In internal testing, users preferred Solaris-rendered interfaces over pages coded by frontier LLMs in a majority of matchups, though the system still struggles with text legibility and long session drift.

Runway · Generative AI / Agentic Interfaces · Relevance: 0.8 · Source →

generative-AI, video-AI, no-code, UI-generation, world-model, agentic-interfaces, frontier-research

WATCHAI Healthcare Tool Spots Heart Disease from ECG in Two Seconds

Imperial College London researchers developed an AI model trained on 10.6 million ECGs that can detect heart failure and valve disease — previously requiring months-long ultrasound queues — within two seconds, with 81% accuracy for heart failure and 90% for valve disease. A 590-patient trial across six hospitals is underway, with the goal of NHS deployment within two years.

The Guardian · AI-Driven Decision Support / Healthcare Analytics · Relevance: 0.8 · Source →

healthcare-AI, diagnostic-AI, ECG, predictive-analytics, medical-decision-support, clinical-AI