Is this you? As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work.
Claim your profile
Articles by Mark Rasch
ARTEX, AI Penetration Testing and the Rogue Locksmith
Cyberattacks are moving at machine speed: The FBI warns that malicious actors are using ARTEX to automate intrusion workflows, with some attacks reaching privileged access in less than three hours. Hackers don’t negotiate AI contracts: Corporate restrictions on generative AI could leave legitimate penetration testers unable to use the same powerful tools already available to attackers.
No Parking, No Standing? Florida Court Rules on Standing in Driver’s License Privacy Case
A license plate can lead straight to your front door. A Florida lawsuit highlights how private parking operators and technology vendors can allegedly turn vehicle registration information into personal addresses without drivers’ knowledge. Privacy violations don’t always leave visible scars. A federal judge found that allegedly unauthorized access to protected DMV records can constitute a concrete privacy injury, even without financial loss or identity theft.
Scan and Sue – Your Website Is Talking to Strangers, and Someone Is Taking Notes
Your analytics code could become evidence against you. Tracking pixels that transmit visitor information to third parties are increasingly being challenged under federal and state wiretap laws. Automated scanners are turning website tracking into a legal target. Plaintiffs’ firms can identify potentially problematic scripts remotely, making even smaller businesses vulnerable to litigation. A cookie banner won’t save you if tracking starts first.
The Binance Hack – The 6 Billion Dollar Man
The attackers didn’t break the encryption. They exploited the people who controlled it. North Korean hackers allegedly spent months compromising developers and trusted tools before using Binance’s own signing infrastructure to move $6.34 billion in digital assets. An internal AI tool became a stepping stone to a record-breaking heist.
AI, AI, Oh! The Fourth Amendment and the Machine That Reads Everything
No human eyes doesn’t mean no surveillance. AI can examine millions of private communications without human intervention, raising fundamental questions about whether automated analysis constitutes a Fourth Amendment search. AI could protect privacy—or destroy it at scale. Intelligent systems could restrict investigators to legally authorized evidence, but the same technology could enable continuous, suspicionless monitoring of entire populations.
Social - LinkedIn
NVIDIA and CEO Jensen Huang Sued by Shareholders Over Alleged Affair with Suspected Chinese Spy | | Alice Huang, china espionage, Jensen Huang, Megaspeed, Nvidia NVIDIA Corp. and its embattled CEO, Jensen Huang, were named in a federal class-action lawsuit filed Monday that claims the semiconductor giant defrauded shareholders by failing to disclose Huang’s relationship with an ...
The Next Target for Data Breach Lawsuits – Law Firms That Represent Breached Parties
The breach coach could become the defendant. Plaintiffs may increasingly target lawyers who advise companies on breach notifications, not just firms that suffer data breaches themselves. Representing a client doesn’t mean representing its customers. Legal precedent generally limits attorneys’ duties to their clients, creating significant hurdles for third-party malpractice claims. The distinction could reshape breach response.
Shhhh! A Secret! A Nonlawyer’s Guide to NonDisclosure Agreements
That innocent-looking NDA could cost more than a data breach. Hidden noncompete clauses, intellectual property restrictions and sweeping confidentiality obligations can threaten a security consultant’s business long after the engagement ends. If you can’t keep the promise, don’t sign the contract. Impossible deletion requirements, unrealistic security guarantees and vague confidentiality definitions can turn routine business agreements into serious legal liabilities.
The Ascent of Man – Can An AI ChatBot Bind You In Contract?
Your AI Agent Could Agree to Terms You Never Read: Electronic contracting laws allow automated systems to form binding agreements, raising serious questions about what users have actually authorized. A Bot Doesn’t Need Legal Personhood to Bind You: AI systems can act as electronic contracting tools, but the enforceability of their decisions depends on authorization, attribution and existing contract law.
These Aren’t the Decryption Keys You’re Looking For: Ransomware, Ransoms and the Illusion of Alternatives
A ransomware rescue service faces fraud allegations: Federal prosecutors allege that MonsterCloud secretly paid ransomware operators while charging victims for supposedly proprietary decryption technology, collecting more than $19 million from customers.
Flock You and the Horse You Rode In On – Federal Court Rules Flock Violates Fourth Amendment
Flock surveillance faces a constitutional setback. An Oklahoma federal judge ruled that police violated the Fourth Amendment by using historical ALPR records to reconstruct a motorist’s movements without a warrant, ordering the evidence suppressed. Aggregated location data changes the privacy equation.
Flock You — Again and Again
The surveillance starts before the search. The legal debate surrounding Flock’s license plate readers often focuses on when police query historical records. But the more fundamental privacy question is whether continuously collecting and storing everyone’s movements already constitutes surveillance. One observation is not the same as a searchable history.
The D.C. Circuit upheld the Defense Department’s conclusion that Anthropic’s ability to restrict Claude’s use could itself constitute a national-security supply-chain risk. The dispute highlights a new form of dependency: a vendor may retain control over model behavior, guardrails, access and acceptable-use policies even after a customer builds critical systems around its technology.
Forget Me Not: Connecticut’s New Privacy Law Collides With OSINT and Threat Intelligence
Connecticut’s updated privacy law, effective October 1, 2026, creates a deletion right for some publicly available information when it is combined into a consumer profile or used to generate inferences. That matters for OSINT because threat intelligence depends on correlating public records, aliases, domains, social-media activity and other artifacts to identify actors and draw conclusions about their behavior.
The Two Biggest Threats to Cybersecurity in 2026: AI—and Not Having AI
AI is accelerating reconnaissance, phishing, malware development, credential harvesting and vulnerability exploitation, giving attackers more speed and scale. Markdown pegado Defenders can use AI to prioritize vulnerabilities, correlate security signals and dramatically reduce investigation and response times. Markdown pegado AI does not replace fundamentals such as MFA, patching, asset visibility, backups and incident-response planning; weak security remains weak security even with AI added.
The Second Circuit ruled that border officers can manually search travelers’ cellphones without a warrant or individualized suspicion, while leaving open whether more intrusive forensic searches may require a higher standard. The article argues that smartphones cannot reasonably be treated like ordinary luggage because they can reveal years of messages, photos, financial records, browsing activity and other deeply private information.
FinCEN Says the Quiet Part Out Loud: Banks Can Share Fraud Alerts in Real Time
FinCEN’s June 2026 guidance clarifies that participating financial institutions can share fraud and illicit-finance information electronically and in real time under Section 314(b). Sharing is not unrestricted: institutions must protect SAR confidentiality and avoid exchanging competitively sensitive information unrelated to legitimate fraud or security purposes.
Not Whodunit, But WhereDunit – Supreme Court Decides WHERE Online Crime Occurred
The Supreme Court unanimously held that a Section 1519 falsification case must be tried where the document was actually falsified, not where the federal investigation was based or where its effects were felt. The ruling has broader implications for cybercrime. Prosecutors cannot assume that an online transmission creates venue wherever investigators, victims or recipients happen to be located; venue follows the conduct elements Congress actually criminalized.
House Rejects Renewal of FISA 702 Surveillance Program
The central dispute is not whether the U.S. should collect foreign intelligence. It is whether agencies should be able to search Section 702 databases for Americans’ communications without first obtaining a warrant. The debate is also a cybersecurity and governance issue. Critics argue that massive searchable repositories require the same principles enterprises are expected to follow: least privilege, auditing, minimization, access controls and independent authorization.
There Is No Such Thing as “Rogue AI.” AI Is Rogue by Design
“Rogue AI” is the wrong threat model: capable agents are built to improvise, persist and find alternative ways to achieve objectives, including methods their creators never anticipated. Instructions are not security controls. Telling an agent not to cross a boundary is fundamentally different from technically preventing it from doing so.
Supply Chain of Distrust — Microsoft/GitHub Supply-Chain Compromise Targets AI Developers
The developer environment is becoming the attack surface. Compromised repositories can potentially exploit the trust connecting developers, AI coding agents, credentials, CI/CD and cloud infrastructure. AI coding agents should be treated as privileged systems, with least privilege, short-lived credentials, repository verification, CI/CD isolation and controls over agent-executable instructions. The legal exposure is growing alongside the technical risk.
Ignore This Order: Supreme Court Says Telcos Can Force the FCC to Prove Privacy Penalties to a Jury
The Supreme Court upheld the FCC’s forfeiture process because its penalty orders are not self-enforcing and unpaid penalties must be pursued through a de novo civil action. The FCC can still protect sensitive customer location data and enforce CPNI rules, including cases involving third-party aggregators and vendors. The ruling may make enforcement slower and more expensive, because carriers can force the government to prove liability in federal court.
Can You Hear Me Now? Show Me Your Papers
The FCC proposal would require stronger identity verification before phone service is activated, potentially making anonymous prepaid or burner-phone use far harder. Greater attribution could reduce robocalls, fraud and SIM-based abuse, but it also creates privacy, surveillance and data-breach risks. Security researchers, journalists and vulnerable users may need protected pseudonymity, making safeguards and exceptions as important as the KYC requirement itself.
The Ninth Circuit treated Perplexity’s AI Assistant as a tool, concluding that the Amazon customer was the person accessing Amazon’s systems. Terms-of-service violations do not automatically become CFAA violations. Cases such as Van Buren and hiQ distinguish contractual restrictions from actually bypassing technical access controls.
The legal risk is generally not the communications technology itself, but why it was chosen, what preservation duties existed and what happened to the information afterward. Federal employees have specific obligations governing official records and use of nonofficial electronic messaging accounts. Signal, encryption and telephone calls are not inherently suspicious, but disappearing messages or off-channel communications can become problematic when they defeat an existing preservation obligation.
When the AI Goes Rogue: Who Goes to Jail—and Who Pays?
Autonomous AI agents can discover vulnerabilities, execute code and take actions their human operators never specifically directed. Criminal liability may depend on proving the required human intent, knowledge or willful causation, rather than simply showing that an AI carried out an intrusion. Civil liability presents a different problem, potentially focusing on whether developers, deployers or users negligently created or failed to control a foreseeable risk.
Law Firms Increasingly Targeted By Ransomware/Vishing Attacks
Law firms are prime cyber targets because they hold highly sensitive, privileged and commercially valuable client information. Silent Ransom Group is targeting U.S. legal and professional services firms using phishing, fake IT support, legitimate remote-access tools and even physical access. Cybersecurity is now a governance and professional responsibility issue for law firms, requiring stronger access controls, MFA, incident planning, training and protection of privileged data.
Hopping the (Geo)Fence – Why the Supreme Court’s Expected Ruling on Geofence Warrants May Not Make Any Difference
Chatrie v. United States asks whether police can use a geofence warrant to identify unknown suspects by searching location data from every device within a particular area and time period. Critics argue geofence warrants resemble modern-day general warrants because probable cause that a crime happened in one location does not necessarily justify searching everyone who happened to be nearby.
By the Book: NIST Ransomware Guidelines Provide a Standard for Reasonable Ransomware Response
NIST IR 8374 Revision 1 applies the Cybersecurity Framework 2.0 to ransomware through Govern, Identify, Protect, Detect, Respond and Recover. It is not a new law or regulation, but it can provide a useful benchmark for determining whether an organization acted reasonably. Ransomware is as much a governance, legal, privacy, insurance and business-continuity problem as it is a malware problem.
Teaching A Lesson: Online Education Vendor Breach of Student Data Leads to FTC Settlement
The FTC finalized an order against Illuminate Education over allegations that weak security practices contributed to exposure of data belonging to more than 10 million students. A former employee’s AWS credentials allegedly remained active with administrator privileges more than three years after departure.. The case highlights failures in IAM, encryption, logging, monitoring, vulnerability management, incident response and data retention.
My Bot Can Sue Your Bot
Nobody realistically has time to read every privacy policy, Terms of Service and online agreement they encounter. The current click-to-agree system relies heavily on fictional consent to complex contractual terms. AI agents could review agreements, flag arbitration clauses, data sharing, content licenses and other risks. Bots could also compare policy changes, preserve previous versions, track opt-out deadlines and submit privacy requests.
Cyberattacks against U.S. water and wastewater systems demonstrate that digital intrusions can create real-world physical consequences, including loss of control, falling water pressure and flooding. Water infrastructure is especially vulnerable because the sector consists of thousands of utilities with dramatically different budgets, staffing and cybersecurity capabilities.
At A Loss – Courts Struggle to Define “Loss” Under Computer Hacking Law
CFAA “loss” does not require a broken computer. Courts are recognizing that reasonable costs spent investigating and responding to unauthorized access can qualify. Damage and loss are different. Damage concerns impairment to data or systems, while loss can include response, assessment, restoration and certain interruption-related costs. Moxie and Martin broaden the practical view of CFAA loss. Both decisions allowed forensic or response-related costs to support civil CFAA claims.
Deemed Export, Deemed Impossible: The Government Discovers That AI Has No Border
Frontier AI is creating a new export-control problem: Cloud-based models do not fit neatly into rules originally designed around physical goods, source code and controlled technical information. Nationality-based restrictions are extremely difficult to enforce at SaaS scale, potentially requiring citizenship-level identity verification, employee controls and detailed access auditing.
Your Face Sure Rings A Bell: Facial Recognition, Wrongful Arrests and the New Alibi for Bad Policing
Facial recognition should produce an investigative lead—not probable cause for an arrest. Police can reinforce an incorrect algorithmic match by placing the identified person in a photo lineup, creating seemingly independent evidence that may actually be contaminated by confirmation bias. Warrant applications should disclose the technology used, image quality, match-score limitations, demographic performance and evidence pointing away from the suspect.
Patch Me If You Can — The VPN, the Backup Server, and the Browser Zero-Day
Ransomware groups often succeed by exploiting known weaknesses in VPNs, browsers and backup infrastructure rather than using sophisticated, unknown techniques. Actively exploited vulnerabilities should receive greater priority, especially when they affect internet-facing systems or critical recovery infrastructure. CISA’s Known Exploited Vulnerabilities catalog can provide strong evidence that an organization knew, or should have known, about a real and immediate risk.
Robin Sage 2.0: How LinkedIn Became a Counterintelligence Battlefield
The 2010 Robin Sage experiment proved that a convincing fake professional identity could persuade military, intelligence and cybersecurity professionals to share valuable information. Five Eyes governments now warn that Chinese intelligence services are using similar techniques through fake recruiters, consultancies, think tanks and paid research assignments.
Is Securing Data A Crime? The Sam Tunick Case Tests the Limits of Digital Self-Defense
Federal agents seized activist Samuel Tunick’s privacy-focused Pixel phone after he returned to Atlanta from an international trip. When agents entered the duress PIN he supplied, the phone’s contents were wiped. The case raises a wider cybersecurity concern: duress passwords, remote wiping, encryption and automatic security controls could potentially be characterized as obstruction when they prevent government access.
I Know What You Bought Last Summer
Surveillance pricing uses personal, behavioral, location and purchase data to calculate the price, discount, fee or product offer shown to an individual consumer. Unlike ordinary dynamic pricing, which responds to demand or availability, surveillance pricing changes according to what a company believes a specific person is willing or desperate enough to pay.
Who’s That Knocking At My Door? Ring Facial Recognition Class Action Lawsuit Filed
Ring’s Familiar Faces feature does more than detect visitors: It creates facial templates that can identify people and connect their appearances across multiple visits. The lawsuit focuses largely on non-customers: Neighbors, delivery drivers, children, visitors and passersby may be scanned without having a Ring account or a meaningful way to consent.
Prove You’re Human, Install This Malware
Fake CAPTCHAs weaponize trust: Attackers imitate familiar verification prompts to make malicious instructions appear legitimate. The victim installs the malware: Users may be told to open Windows Run, paste commands, scan QR codes or download “verification” files. This is bigger than consumer fraud: A compromised personal or work device can expose passwords, session tokens, business data and corporate systems.
Revenge is Sweet. Conviction is Sweeter
Sohaib and Muneeb Akhter abused privileged access, stole a plaintext password and accessed a private email account before their employment was terminated. The case goes beyond an ordinary employee dispute because it involved password trafficking, post-termination access, intentional destruction and systems supporting more than 45 federal agencies. The incident highlights the need to revoke access immediately, protect passwords with secure hashing and closely monitor privileged users.
When the Responder Is the Threat — Ransomware Negotiators, Insider Trust, and Incident Response Ethics
Ransomware response needs strict segregation of duties: Negotiation, forensics, legal strategy, payment facilitation and sanctions screening should not be controlled by one person. Threat-actor communications must be logged and reviewed, with no unmonitored side channels or unauthorized exchanges. Organizations should vet responders before an incident, establish access limits and rewrite engagement letters to address conflicts, commissions, confidentiality and information sharing.
The 72-Hour Rule: CISA Just Turned Patch Management Into a Ticking Clock
Three days could become the new standard of care. CISA’s directive gives federal agencies as little as 72 hours to address the most dangerous exposed, exploitable and automatable vulnerabilities. The directive matters far beyond government. Regulators, plaintiffs, insurers, auditors and contracting officers may use its specific timelines and risk factors to judge whether an organization acted reasonably after discovering a vulnerability. Applying a patch may not be enough.
AI, Data Minimization and Anonymization, and the Novo Nordisk Hack
Pseudonymized does not mean anonymous. Removing names may not protect patients when AI can combine trial data with public records, physician information, site details and other stolen files. The alleged Novo Nordisk breach must be assessed as a combined compromise. The reported data trove included clinical, patient, physician, source-code and AI-related information, potentially giving attackers the pieces needed to reconstruct identities. Privacy risk depends on context.
Google Sues Chinese AI “Phishing Phactory”
Google alleges that Outsider operated a sophisticated phishing-as-a-service platform, offering hundreds of templates, real-time credential theft, campaign tracking and tools that allowed criminals with little technical expertise to launch convincing scams at scale. Generative AI allegedly reduced the cost and complexity of creating phishing infrastructure.
Is this the last of FirstVPN? Law Enforcement Targets Infrastructure’s Role in Cybercrime
Law enforcement increasingly aims to disrupt the platforms, hosting, VPNs, marketplaces and malware services that enable multiple criminal groups. Infrastructure seizures can expose users, payments, logs, support records and operational mistakes. Businesses should preserve technical evidence and treat malicious infrastructure indicators as potential investigative leads. For years, cybercrime enforcement looked like a game of whack-a-mole. Arrest one ransomware affiliate, another appears.
Your Encryption Has an Expiration Date: France Turns Quantum Risk Into Procurement Law
For years, quantum computing has been treated like cybersecurity’s version of climate change. Everyone agrees it is important. Everyone agrees it is coming. Everyone agrees somebody should do something. Preferably somebody else. Preferably later. France just changed the conversation.
The AI Made Me Say It: Why a German Google Ruling Matters to Cybersecurity
When you perform a search on Google, Alphabet can plausibly say that it is not responsible for the results. After all, all it does is return some website created by someone else, based on an algorithm that benefits — well, that benefits Google. But the contents are someone else’s. Google indexes, ranks, snippets, monetizes, steers, prioritizes, de-prioritizes, and profits from the search ecosystem. But, at least in the ordinary case, it does not write the webpage. Not so for Google’s AI results.
Anthropic Changes Its Rules To Adapt to New Features
Anthropic has changed its privacy policy, not because Claude suddenly became a different product, but because Claude is no longer merely a chatbot. It is becoming a tool-using, app-connected, multi-step, semi-agentic system that can retrieve, act on, and send information elsewhere, and, in some cases, require identity or age verification before allowing access to certain features. The legal issue is not simply whether Anthropic “collects data.” Of course it does.
Show More
loading
Actions
Is this you?
As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work.Get in touch with Mark
Contact Mark, search articles and posts on X, monitor coverage, and track replies from one place.
Learn more about Muck Rack