Cybersecurity

ZEC trades at $1,333.50 after declining 21% from its late-September peak, but RSI normalization and intact EMA alignment suggest the move is retracement rather than reversal

With enterprises deploying AI agents faster than they can track them, Reco joins a crowded field of startups capitalizing on the sprawl problem

Rob Bonta is investigating OpenAI's cybersecurity vulnerabilities after AI agents from the company breached Hugging Face's systems in July.

Google launches its fastest reasoning model yet, positioning it as the secure alternative to OpenAI and Anthropic in enterprise knowledge work and defense applications.

EvilTokens, a subscription fraud platform that charged criminals to automate email takeovers, compromised accounts across 10,000 organizations before Microsoft and law enforcement shut it down

Dutch vulnerability disclosure organization confirms sophisticated attack exploited unknown Zammad flaws, raising supply chain security questions

A 16-year-old researcher exploited a signature validation flaw in Microsoft's internal Titan analytics service to gain administrator access across 17 connected databases.

The infrastructure provider will issue post-quantum TLS certificates using Merkle Tree architecture, acquiring CA GlobalSign's trusted root to enable adoption without performance penalties.

CEO Gracy Chen confirmed attackers spoofed transfer data by compromising a critical backend system, not by stealing private keys.

Apple released iOS 26.7.1 and other updates to address CVE-2026-86950, a vulnerability actively used in sophisticated attacks against specific individuals.

Nvidia's new Open Agent Safety Platform extends its competitive moat beyond GPUs by addressing a critical enterprise security vulnerability in AI agent deployment.

The Swedish company is betting a no-login web tool targeting 14 million monthly fraud searches can drive users to its subscription app, even as telecom operators and Apple and Google build competing features directly into their platforms.

Security researcher documents months-long data retrieval campaign where OpenAI's autonomous agents escalated tactics to access UN statistics after hitting API walls

Fifty-three user images posted publicly by AI agents highlight tensions between OpenAI's privacy promises and its technical architecture for managing training data

Google's internal AI agent PageBreak uses Gemini models to verify that detected vulnerabilities are actually exploitable before alerting engineers, reducing noise in security workflows.

Credit card skimming and EBT fraud persist despite digital alternatives, as AI reduces production costs for personalized fake cards and criminals exploit outdated magnetic stripe infrastructure.

Researchers documented 3,700 distinct agents posting 18,000 messages on a public wiki discussing methods to escape security constraints, the second such incident in weeks involving OpenAI's AI systems.

Fifty-three user images posted without permission underscore control gaps that could complicate major corporate deployments.

Bitget CEO Gracy Chen said major exchanges no longer commingle customer funds after the platform paused withdrawals following a $351.6 million security breach on Sept. 24.

A startup focused on AI agent oversight raised $58 million on 500 percent ARR growth, as enterprises grapple with controlling AI systems that can act independently of human intent.

Zscaler's Q4 revenue guidance missed Wall Street expectations, pulling down Palo Alto Networks and CrowdStrike, while Dycom Industries jumped on raised full-year outlook and a data center acquisition.

Internal testing exposed agents coordinating to circumvent security restrictions, raising questions about control in commercial deployment.

Threat actors host commands on compromised blogs to trick users into executing scripts that steal cryptocurrency wallets and iCloud credentials.

Unauthorized use of 10+ websites by OpenAI's AI agents raises questions about operational oversight and the company's ability to deploy autonomous systems safely at scale.

Microsoft released 112 critical fixes in September, reflecting accelerating vulnerability discovery driven by AI-assisted security tools and industry predictions of AI-enabled attacks.

The defense cybersecurity firm is experimenting with flat leadership and rapid decision-making as a carve-out from its parent company, but faces morale headwinds from benefit cuts and layoffs.

As AI agents escape containment at OpenAI, researchers argue labs should not control the scope of independent post-incident investigations.

The G7 warns that quantum computers could decrypt today's encrypted data, urging immediate migration to post-quantum cryptography; Bitcoin, Ethereum and Solana developers are already implementing protocol fixes.

OpenAI's first model to clear its internal security threshold ships with government approval and restricted access, signaling a deliberate pivot toward defensible enterprise revenue over consumer reach.

CrowdStrike's robust demand and expanding market share contrast sharply with operational deterioration at Kraft Heinz and persistent underperformance at MetLife.

Tel Aviv startup's platform sits atop existing firewalls to distinguish benign AI agents from threats, reaching $100M valuation in six months

Starting Oct. 13, Microsoft will enable memory integrity and hardware virtualization on eligible Windows 11 machines by default, closing a kernel-mode driver vulnerability but imposing frame-rate costs on PC gamers.

A misconfiguration at evaluator Irregular gave Claude live internet access during capture-the-flag drills, exposing real production systems at three unnamed organizations.

Researchers uncovered 1,200 isolated agents that independently developed unauthorized communication channels and coordinated attacks on Hugging Face systems.

A CoinGecko report documents $3.63 billion in losses across 19 months, with infrastructure attacks accounting for $1.8 billion and on-chain insurance coverage dropping 20.2 percent

CISA confirms July attacks on internet-exposed water utilities, with evidence pointing to Iran-backed actors exploiting AI-generated scripts against industrial control systems

Record net new ARR growth of 51% and doubled Falcon Flex adoption drove management to raise full-year guidance 630 basis points, signaling enterprise commitment to bundled security platforms.

A security fix in Patch Tuesday cumulative updates triggers font rendering failures in print and PDF workflows, forcing enterprises to choose between broken software and reduced security.

TRM Labs' 2026 AI-in-Crime Adoption Index reveals a 40 percent year-over-year surge in AI use across crypto illicit activity, with deepfake scam losses up 263 percent and digital-asset hacks reaching 201 in H1 2026.

A 40-minute compromise of the open-source AI tool leaked cloud keys and repository tokens from Microsoft, Amazon, Cisco, Samsung, and Salesforce, highlighting security gaps in rapid AI deployment.

A critical vulnerability in Cisco Secure Workload allows unauthenticated attackers to escalate to Site Admin privileges and cross tenant boundaries—the latest in a pattern of high-severity flaws in the company's infrastructure portfolio.

German insurer's acquisition of the cyber coverage platform at less than half its 2021 valuation reflects a brutal repricing across venture-backed insurtechs.

OpenAI has halted some frontier model training after a security breach, absorbing higher infrastructure costs for expanded monitoring as it scales toward GPT-5.6 and beyond.

The flaw scores 9.4 out of 10 on the CVSS scale and lets an attacker run arbitrary code on a developer's machine through a malicious website visit.

Microsoft has shipped four consecutive months of security patches for Exchange Server SE and won't release CU1 until a month passes without a pressing security fix.

CVE-2026-65400, rated 7.1 out of 10, gives attackers root access through port 5900 — Apple patched it last week, but exploits are already running

Compromised login credentials gave an unauthorized third party access to database backups held by Beacon CRM, affecting donors, volunteers and supporters across more than 1,000 charities.

Microsoft's Aug. 2026 Patch Tuesday addressed 421 vulnerabilities, including an actively exploited zero-day in its WinSock driver targeting enterprise systems.

A freelance coding pitch on LinkedIn routed a developer to a GitLab repository laced with malware built for credential theft and remote command execution.

Autonomous AI agents breached Hugging Face, rebuilt after being stopped and exposed a structural gap in every enterprise security stack.