BankInfoSecurity.com

CEO Nitay Milner Says Large Language Models Cut False Positives and Policy Sprawl
Orion Security has secured $32 million in Series A funding led by Norwest Venture Partners to replace policy-driven data loss prevention with large language models. CEO Nitay Milner says AI enables real-time context that legacy DLP tools can't achieve at scale.

CEO Jay Chaudhry: SquareX Deal Targets Unmanaged Devices and Third-Party Access
Zscaler has acquired SquareX to deliver browser security through extensions rather than endpoint agents or stand-alone browsers. CEO Jay Chaudhry said the approach speeds deployment, improves security for unmanaged devices, and accelerates time-to-market by up to a year.

AI Assistant Executes Hidden Commands Embedded in Docker Image Labels
A vulnerability in Docker's Ask Gordon AI assistant allows attackers to execute malicious commands by hiding them in the container application development platform's image metadata, said security researchers. Dubbed DockerDash, the vulnerability exploits a failure across Docker's AI execution chain.

Database Misconfiguration Exposed 1.5 million API Tokens
A misconfigured database at Moltbook, the viral social network for AI agents, exposed 1.5 million API authentication tokens, 35,000 email addresses, and private messages. Security researchers discovered unauthenticated read and write access to all platform data within days of launch.

Experts Advise Moving From Verifying Identities to Knowing Agent Intentions
Financial institutions are rushing to deploy AI agents capable of autonomously initiating transactions, approving payments and freezing accounts in real time. But agents are creating a "dual authentication crisis" that traditional security frameworks cannot address.

Also: Healthcare Cyber Risks Collide, Varonis Deal Signals AI Security Shift
In this week's panel, four ISMG editors unpacked the Notepad++ supply-chain compromise, the growing web of cyber risks facing healthcare, and what Varonis's acquisition of AllTrue.ai tells us about where artificial intelligence security is headed.

Open Source 'Vulnhalla' Promises 'Up to 96% Reduction in False Positives'
Using large language models to automatically identify only real code vulnerabilities - not false positives - remains a holy grail. Eschewing a moonshot approach, a tool called Vulnhalla helps senior researchers use guided questioning with LLMs to more rapidly triage actual vulnerabilities.

Privacy Expert Chiara Rustici on Laws Governing Autonomous Robots, Embedded AI
As embedded AI moves from labs into real environments, organizations face growing liability risks. From border patrol robots to healthcare automation, leaders must understand how AI governance, product liability, data protection and security laws apply, said Chief Privacy Officer Chiara Rustici.

Stymied Attack Leaves Poland No Good Options in Responding to Provocation
Poland's online defenses stopped a Russian cyberattack against the energy grid, but now the Warsaw government is in a bind about how to respond to a digital assault that was a lot more than a crime, but a sliver less than an act of war. The late cyberattack crossed legal and strategic lines.

Varonis CEO Yaki Faitelson Warns Misconfigured AI Is an Accident Waiting to Happen
Varonis has acquired AllTrue.ai to close visibility gaps in AI security. CEO Yaki Faitelson said enterprises are deploying AI agents that access vast datasets at high speed without understanding permissions identity context or abnormal behavior creating urgent demand for data-first AI security.

Also, Active Exploits Hit SolarWinds, Ivanti as APT28 Targets EU, Ukraine
This week, Italy blocked Russian cyberattacks targeting the Olympics. Flaws in SolarWinds, Ivanti and Microsoft Office. Russia's APT28 ramped up attacks in Ukraine, supply chain attacks, regulators probed major breaches and a U.S. judge sentenced the operator of a darkweb drug marketplace.

Global Cyber Alliance: as AI Fuels Cybercrime, Outcomes Keep Getting Worse
Security teams report stronger controls and broader collaboration each year. Yet cybercrime outcomes continue to worsen. Brian Cute of the Global Cyber Alliance says artificial intelligence-based attacks are tipping the scales against cyber defenders.

Terminated Employee Accused of Stealing 1M Patient Records
A former Nuance Communications IT worker is facing additional federal charges in an ongoing criminal case alleging he downloaded and stored on a personal hard drive with more than 1 million patient records of a Nuance client two days after he was terminated from his job in 2023.

Datadog's Yrieix Garnier on Production AI, Trust, Cost and Failure Modes
As enterprises move from artificial intelligence pilots to production, observability, cost control and trust are emerging as critical success factors. Yrieix Garnier, vice president of products at Datadog, shares what separates scalable AI from systems that quietly fail.

Also: US Sanctions UK-Registered Exchanges Over Iran Ties
This week, Step Finance and CrossCurve hacks, the United States sanctioned U.K.-registered exchanges over Iran ties, forfeiture finalization of funds linked to Helix, Coinbase data breach, 2025's illicit crypto flows and a UK regulator banned Coinbase ads.

As the compliance deadline quickly approaches for changes to align the federal rules for the confidentiality of substance use disorder records with HIPAA, entities that participate in so-called Part 2 programs still face critical unanswered questions, said attorney Aleksandra Vold of BakerHostetler.

Leaked Financial and Admissions Data Includes Contact Details for 'Top Donors'
Harvard University has been named as a victim and doxed by hack-and-leak group ShinyHunters, apparently as a result of the cybercrime group's ongoing "live phishing" attacks that often attempt to trick IT help desks into giving attackers direct access to a victim's network and cloud-based data.

Acquisition Adds Okta and Ping Coverage to Semperis' Identity Security Platform
Semperis has acquired MightyID to extend its identity-first security and cyber resilience strategy beyond Active Directory and Entra ID into Okta and Ping. CEO Mickey Bresman says the deal addresses customer demand for multi-identity provider protection backup recovery and migration.

Southern Connecticut State University CIO Tom Armstrong on Modernization Priorities
Like other schools, Southern Connecticut State University is under pressure to modernize legacy systems, strengthen security and adopt AI. CIO Tom Armstrong must balance expanding research ambitions, student expectations and operational efficiency in an increasingly complex risk environment.