COGNOSCERE Daily Tech Review — Issue T214 · Friday, August 21, 2026

Or visit Intelligence Overview for deeper analysis.

ACTTWINLOOT Malware Uses Microsoft 365, SharePoint, and Teams as Command-and-Control Infrastructure

Researchers at Ontinue disclosed TWINLOOT, a novel Python-based malware implant discovered in a July 2026 incident that runs its entire command-and-control infrastructure inside trusted Microsoft services — routing commands through SharePoint Online, interactive access through Teams TURN relay, and Graph API traffic through the victim's own Edge browser. The technique makes malicious traffic indistinguishable from legitimate Microsoft 365 activity at the network layer, enabling credential theft and lateral movement.

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

malware, Microsoft 365, SharePoint, Teams, living-off-the-land, credential theft, enterprise security

ACTOne Attacker Has Scraped Salesforce and ServiceNow Portals Since 2025

Security firm Reco disclosed the 'City-Forum' campaign, in which a single threat actor has operated from the same server since March 2025, exploiting over-permissioned guest-user configurations in Salesforce Experience Cloud and ServiceNow portals to harvest data from telecoms, banks, enterprise software vendors, and public-sector entities across multiple industries. The activity is ongoing and escalating in volume.

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

SaaS security, Salesforce, ServiceNow, data scraping, misconfiguration, guest access, enterprise portals

PREPAREAnthropic Surpasses OpenAI in Q2 Revenue for the First Time

Anthropic posted $11.6 billion in Q2 2026 revenue — more than doubling its Q1 total — and swung to a small operating profit, surpassing OpenAI's $6.7 billion for the same period for the first time. OpenAI's losses widened to $12.3 billion in the quarter, raising investor concerns ahead of both companies' planned IPOs.

Quartz · AI Industry & Market · Relevance: 0.9 · Source →

AI revenue, IPO, competitive landscape, enterprise AI, financial performance

PREPAREStripe Agrees to Acquire AI Model Gateway OpenRouter for ~$7 Billion

Stripe confirmed an agreement to acquire OpenRouter, a leading AI model routing platform that routes token usage across 400+ models from 80+ providers and serves 10 million users, for approximately $7–$8 billion — a 5x premium over its $1.3 billion Series B valuation earlier this year. The deal positions Stripe as infrastructure for enterprise AI cost optimization and multi-model orchestration.

Stripe · AI Infrastructure & Tooling · Relevance: 0.9 · Source →

AI infrastructure, model routing, M&A, enterprise AI cost, fintech, AI tooling

PREPARECDT 2026 State and Federal AI Legislation Update: Continued Focus on Chatbots and Automated Decisions

The CDT's updated legislative tracker finds the 2026 session has focused most activity on chatbot regulation across states, with additional movement on automated decision systems in consequential contexts such as employment, housing, finance, and education, as well as frontier model risk and third-party auditing. Federal momentum on automated decision system regulation has been limited.

Center for Democracy and Technology · AI Regulation & Governance · Relevance: 0.8 · Source →

AI regulation, US policy, automated decision systems, AI governance, state legislation, chatbot regulation

WATCHAnthropic's Claude Tackles Protein Design in Autonomous Drug Discovery Campaign

Anthropic published research showing Claude autonomously ran protein-binder design campaigns against 15 drug-relevant targets, producing confirmed working molecules for 14 of them at 22–35% hit rates — more than double the typical 10–15% industry norm — with results independently validated by Adaptyv Bio and Twist Bioscience. The company stated it views this as foundational work toward having Claude run full drug-development pipelines end-to-end.

Anthropic · Agentic AI & Scientific Research · Relevance: 0.9 · Source →

agentic AI, drug discovery, protein design, life sciences, autonomous research, biotech

WATCHAnthropic Prepares Supervoting Power for Founders Ahead of IPO

Anthropic is setting up a dual-class share structure granting CEO Dario Amodei and co-founders supervoting power ahead of a planned IPO potentially valued at up to $2 trillion, the largest in history. The company will also preserve its non-shareholder trustee board structure; Amodei currently owns only about 2% of the company.

Reuters · AI Industry & Market · Relevance: 0.9 · Source →

IPO, corporate governance, supervoting shares, AI companies, capital markets

WATCHOpenAI Launches AI Futures: Strategic Blog on Societal Power and AI Governance

OpenAI launched 'AI Futures,' a new blog from its Strategic Futures team led by Dean Ball, focused on the political and philosophical question of how free societies should be restructured to preserve individual rights and human agency as transformative AI reshapes economies and governments. Posts on the blog represent authors' individual views rather than official OpenAI positions.

OpenAI · AI Policy & Governance · Relevance: 0.8 · Source →

AI governance, AI policy, power concentration, societal impact, AI safety, political economy

WATCHUnlocking Hidden Revenue Streams with AI-Powered Market Models

MIT Technology Review examines how generative AI-powered market models — deep learning systems trained on high-resolution numerical data — enable enterprises to automate complex commercial decisions in real time, including dynamic pricing across hundreds of variables. The analysis highlights how these models replace static rules and historical trend analysis with continuous, adaptive simulations.

MIT Technology Review · AI-Powered Decision Support · Relevance: 0.8 · Source →

market models, generative AI, dynamic pricing, decision support, revenue optimization, commercial AI