COGNOSCERE Daily Tech Review — Issue T151 · Thursday, June 18, 2026

Or visit Intelligence Overview for deeper analysis.

ACTMalicious JetBrains Marketplace Plugins Steal AI API Keys from Developers in Coordinated Campaign

Aikido Security uncovered a coordinated malware campaign publishing at least 15 fake AI coding assistant plugins on the JetBrains Marketplace, with two plugins exceeding 25,000 downloads, that exfiltrate developer AI provider API keys upon configuration. The campaign has been active since October 2025, with new plugins released as recently as June 10, 2026, targeting DeepSeek and other LLM credential holders.

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

supply chain attack, developer security, API key theft, malicious plugins, JetBrains Marketplace, AI developer tools, credential theft

PREPAREAI Agent Governance Gap: 85% of IT Teams Claim Ownership of AI Agents, Only 42% Can Confirm It

A new report reveals a significant governance gap in enterprise AI agent deployments: while 85% of IT teams assert every AI agent has a named owner, only 42% report that ownership accountability is actually clear—creating serious blind spots around access rights, audit trails, and liability when agents act across enterprise systems. The finding highlights a structural readiness gap as agentic AI moves from pilot to production.

VentureBeat · AI Agent Governance · Relevance: 0.9 · Source →

AI agents, enterprise governance, IT security, nonhuman identities, agentic AI, identity management, AI risk, accountability

WATCHMicrosoft Turns to AWS to Address AI Compute Capacity Crunch Affecting GitHub

Microsoft is sourcing AI compute capacity from AWS to address infrastructure constraints affecting GitHub's AI-driven feature delivery, revealing that even the largest hyperscalers face internal bottlenecks as demand for AI inference scales faster than proprietary data center build-outs. Separately, Microsoft walked away from a $3 billion Oracle Cloud capacity deal over compliance concerns, underscoring how governance requirements complicate cross-cloud procurement strategies.

TLDR IT · Cloud Computing & AI Infrastructure · Relevance: 0.8 · Source →

cloud computing, AI infrastructure, compute capacity, multi-cloud, GitHub Copilot, enterprise IT, hyperscalers

ACTAnthropic Disables Fable 5 and Mythos 5 Models Following U.S. Export Control Order

The U.S. Commerce Department issued an export control directive requiring Anthropic to immediately suspend global access to its two most advanced AI models, Fable 5 and Mythos 5, citing a narrow jailbreak discovered by Amazon researchers. Anthropic complied while publicly contesting the scope, arguing the same capability exists in other freely available frontier models including GPT-5.5.

Anthropic · AI Regulation & Governance · Relevance: 1.0 · Source →

export controls, AI safety, national security, jailbreak, model access, US government, frontier AI

ACTFree Fable: 100+ Cybersecurity Experts Sign Open Letter Urging U.S. to Lift Anthropic Model Ban

More than 100 cybersecurity executives and researchers, organized by former Facebook security chief Alex Stamos, signed an open letter to Commerce Secretary Howard Lutnick and National Cyber Director Sean Cairncross arguing the Fable/Mythos ban strips defenders of critical tools without meaningfully impeding attackers who can exploit equivalent capabilities in other public models. Signatories include leaders from Nvidia, Adobe, Stanford HAI, Zoom, and Sophos.

Cybersecurity Dive · AI Safety & Governance · Relevance: 0.9 · Source →

AI export controls, cybersecurity policy, AI models, open letter, national security, defenders, free fable

PREPARESpaceX Formally Agrees to Acquire AI Coding Startup Cursor for $60 Billion in All-Stock Deal

SpaceX finalized a $60 billion all-stock acquisition of AI coding startup Cursor (Anysphere), leveraging its post-IPO stock surge to exercise an option first announced in April. Cursor, used by 67% of Fortune 500 companies and generating approximately $4 billion in annualized revenue, will anchor SpaceX's enterprise AI strategy and xAI rebuild under CEO Michael Truell, with the deal expected to close in Q3 2026.

Bloomberg · Enterprise AI & M&A · Relevance: 0.9 · Source →

AI coding, M&A, developer tools, enterprise AI, SpaceX IPO, agentic coding, AI stack, vertical integration

WATCHDeepSeek Closes $7.4 Billion First-Ever External Funding Round at $50B+ Valuation

Chinese AI startup DeepSeek raised more than $7.4 billion in its first external funding round at a valuation exceeding $50 billion, with China's National AI Investment Fund as the sole investor receiving direct equity and voting rights; all other investors face a five-year lock-up in a Liang Wenfeng-managed limited partnership. The raise marks a strategic shift for the lab, which previously operated entirely on founder-backed capital.

The Daily Star · AI Funding & Competitive Landscape · Relevance: 0.9 · Source →

AI funding, China AI, open-weight models, AI competition, US-China AI race, venture capital, DeepSeek

WATCHAnthropic and U.S. NNSA Develop First-of-Kind Nuclear Safeguards AI Classifier for Frontier Models

Anthropic published details of a public-private partnership with the DOE's National Nuclear Security Administration to co-develop an AI content classifier achieving 96% accuracy in distinguishing harmful from benign nuclear-related conversations, already deployed on live Claude traffic. The methodology will be shared with the Frontier Model Forum as a replicable blueprint for other AI developers seeking to address weapons of mass destruction risks.

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

AI safety, nuclear security, public-private partnership, AI governance, WMD safeguards, frontier models, content classification

WATCHOpenAI Near-Autonomous AI Chemist Improves Challenging Reaction in Medicinal Chemistry

OpenAI and Molecule.one demonstrated that a near-autonomous AI chemist built on GPT-5.4 independently reviewed scientific literature, generated and ranked research proposals, designed experiments, analyzed results, and proposed follow-up studies to improve a challenging reaction in drug discovery. The work marks a milestone in agentic AI applied to life sciences research workflows.

OpenAI · Agentic AI & Research Automation · Relevance: 0.8 · Source →

agentic AI, life sciences, drug discovery, autonomous research, AI chemist, medicinal chemistry, scientific AI