Transcript
Dave Bittner (0:02)
You're listening to the Cyberwire Network powered by N2K. We've all been there. You realize your business needs to hire someone yesterday. How can you find amazing candidates fast? Well, it's easy. Just use Indeed when it comes to hiring, Indeed is all you need. Stop struggling to get your job post noticed Indeed Sponsored Jobs helps you stand out and hire fast. Your post jumps to the top of search results so the right candidates see it first and it works. Sponsored Jobs on indeed get 45% more applications than non sponsored ones. One of the things I love about Indeed is how fast it makes hiring. And yes, we do actually use Indeed for hiring here at N2K CyberWire. Many of my colleagues here came to us through Indeed plus plus with Sponsored Jobs. There are no subscriptions, no long term contracts. You only pay for results. How fast is Indeed? Oh, in the minute or so that I've been Talking to you, 23 hires were made on Indeed according to Indeed Data Worldwide. There's no need to wait any longer. Speed up your hiring right now with Indeed and listeners to this show will get a $75 sponsored job credit. To get your jobs more visibility at indeed.com cyberwire just go to indee indeed.com cyberwire right now and support our show by saying you heard about Indeed on this podcast. Indeed.com cyberwire terms and conditions apply. Hiring Indeed is all you need. X Twitter had multiple waves of outage yesterday Signals president warns against agentic AI A new lawsuit alleges DOGE bypassed critical security safeguards Is the Five Eyes alliance fraying? The NINJA attack poisons AI memory through user interaction. Researchers report increased activity from the Sidewinder APT group. A critical Veritas vulnerability enables remote code execution. A Kansas healthcare provider breach exposes 220,000 patients Data New York sues Allstate over data exposure and insurance Web CISO warns of critical Avanti and Veracode vulnerabilities. The FTC is going to refund 25 and a half million dollars to victims of tech support scams. On our Industry Voices segment, we're joined by Gerald Bushelt, CISO at Acronis, who's discussing how threat research and intelligence matter to MSPs. And the UK celebrates a record breaking cyber first girls compet foreign March 11, 2025 I'm Dave Bittner and this is your Cyberwire Intel Briefing. Thanks for joining us here today. It is great to have you with us. A cyber attack caused outages on X Twitter on Monday with reports indicating multiple attack waves. While Elon Musk called it a massive cyber attack and suggested a coordinated group or nation state was involved. Details remain unclear. Musk later pointed to IP addresses from Ukraine, but sources say most attack traffic came from the US, Vietnam and Brazil. The attack was likely a DDoS attack where compromised devices overwhelm a system with traffic. The Dark Storm Team, a pro Palestine hacktivist group possibly linked to Russia, claimed responsibility. Other groups, including Anonymous affiliated hacktivists, also took credit, but verifying these claims is difficult. Cyber attacks like these often blur the lines between hacktivism, cybercrime and state sponsored operations. X Twitter has been targeted before, including by Anonymous Sudan, a group whose members were recently charged in the US for offering DDoS services. Investigations into this latest attack are ongoing. Speaking at the south by Southwest conference, Signal President Meredith Whitaker warned that agentic AI poses serious privacy and security risks. She compared AI agents to putting your brain in a jar as they perform tasks on users behalf such as booking tickets, managing calendars and sending messages. To function, these agents would need deep access to user systems including web browsing, credit card details, messaging apps and calendars, likely with route level permissions. She cautioned that processing such tasks would almost certainly happen on cloud servers. Exposing sensitive data Whitaker stressed that integrating AI agents with secure messaging apps like Signal would compromise message privacy. She also criticized the AI industry's reliance on mass data collection, arguing that prioritizing bigger is better AI risks further eroding privacy in exchange for convenience. A new lawsuit alleges the Department of Government Efficiency Doge bypassed critical security safeguards at the Social Security Administration, risking exposure of sensitive data. Former SSA Acting Chief of Staff Tiffany Flick warned that Doge operatives led by Mike Russo, pressured officials to grant system access to Akash Baba despite unresolved security clearances. Doge's push for unrestricted data access ignored federal protections designed to prevent financial exploitation and unauthorized system breaches. Flick accused Doge of forcing staff to share highly sensitive information via potentially unsecured email channels, relying on AI tools to analyze data and determine federal job cuts. She resigned after security policies were disregarded and Leland Dudek, a mid level analyst, was elevated to acting commissioner. The AFL CIO backed lawsuit warns that Doge's actions jeopardize national security, with federal cybersecurity experts sounding alarms over mass government dismissals and weakened data protection measures, NBC News reports Several US allies are reconsidering their intelligence sharing protocols, fearing that President Trump's warming ties with Russia could compromise sensitive data, sources say concerns center on protecting foreign assets as intelligence agencies are bound by strict commitments to shield sources identities Members of the Five Eyes alliance, the uk, Canada, Australia, New Zealand, along with Israel and Saudi Arabia are evaluating whether to limit intelligence flow to Washington. While publicly downplaying concerns, some officials privately question U.S. reliability and the risk of intelligence leaks. Trump's recent pauses in intelligence assistance to Ukraine and the reported halt of cyber operations against Russia have heightened security worries. Some fear a U S Russia cyber detente. Despite Russia's history of harboring cybercriminals, former intelligence officials warn that Moscow is an unreliable partner and scaling back intelligence sharing could undermine global security efforts. Researchers from Michigan State University, University of Georgia, and Singapore Management University have uncovered a new attack method that manipulates AI models with memory without requiring backend access. Dubbed Minja for Memory Injection Attack, the technique allows a regular user to poison an AI's memory simply by interacting with it. The attack injects misleading prompts into the model's memory, altering future responses. Tested on GTP4 powered AI agents, Minja tricked a medical chatbot into swapping patient records, a webshop AI into misdirecting purchases, and a QA agent into answering questions incorrectly. With over 95% injection success, Ninja bypasses traditional moderation filters by disguising manipulations as legitimate reasoning. The findings highlight serious security risks for AI systems with memory, urging immediate improvements in AI memory safeguards. OpenAI has not yet commented on the vulnerability. Researchers at SecureList report increased activity from the Sidewinder APT group in 2024 with enhanced malware, expanded targets, and global reach. Traditionally focused on military and government entities, the group now targets maritime, logistics and nuclear nuclear sectors across South Asia, Southeast Asia, the Middle east, and Africa. Using Spear phishing emails, Sidewinder exploits a vulnerability to deploy Stealer Bot, a post exploitation toolkit. Their malware, disguised as legitimate DLL files, includes advanced evasion techniques like control flow flattening. Sidewinder rapidly adapts modifying malware within five hours of detection. Their continued reliance on old vulnerabilities underscores the importance of patching outdated systems to defend against sophisticated threats targeting critical infrastructure worldwide. A severe remote code execution flaw in Veritas Arcterra Infoscale exposes enterprise disaster recovery infrastructure to attck. The issue stems from insecure deserialization in the Windows plugin host service, allowing attackers to execute arbitrary code via malicious Net remoting messages. The flaw affects infoscale versions 7.0 and 8.0.2 on Windows with system level privilege risks. Veritas advises disabling plugin host or using manual Dr. Configurations to mitigate exposure. Security experts warn that outdated technologies like net deserialization remain prime targets requiring proactive defense. Beyond patching, organizations should audit doctor workflows to prevent exploitation. A December cyber attack on Sunflower Medical Group compromised 221,000 patients Sensitive data including Social Security numbers, medical records and insurance details. The breach, discovered January 7, revealed hackers had been inside the system since mid December, stealing files. While Sunflower has not confirmed a ransomware attack, the Raisida ransomware gang claimed responsibility, demanding $800,000. The company notified regulators, offered credit monitoring and stated no operational disruptions occurred. Raisida has previously targeted healthcare and nonprofit organizations, heightening concerns over medical data security. New York State is suing all state insurance for failing to secure personal data, allowing criminals to steal thousands of driver's license numbers from poorly designed quote generating websites. The issue stemmed from National General, an all state unit which exposed driver's license numbers in plain text during the quoting process. Fraudsters exploited the system, harvesting at least 12,000 records for identity theft and unemployment fraud. The breach went undetected for over two months, with 9,100 New Yorkers affected, yet National General failed to notify them, violating state laws. Another 187,000 individuals data was compromised due to weak access controls, including plain text passwords and no multi factor authentication for insurance agents. New York seeks penalties and an injunction against continued security failures. Texas has also sued Allstate for allegedly collecting telematics data without user consent, further raising privacy concerns. CISA has added three critical Ivanti endpoint management vulnerabilities to its known exploited vulnerabilities catalog. These path traversal flaws allow unauthenticated attackers to leak sensitive information remotely. CISA also flagged two veracode vulnerabilities, an unrestricted file upload flaw and an SQL injection vulnerability. The agency urges all organizations to immediately patch these issues to prevent cyber attacks. The Federal Trade Commission will begin distributing $25.5 million in refunds to over 736,000 consumers deceived by Restoro and Reimage, tech support companies that used fake system warnings to trick users into paying for unnecessary computer repairs. These firms impersonated Windows Pop ups, falsely claiming devices had malware or performance issues. Investigators found their software fabricated security threats to push users into buying repair plans ranging from $58 to $499. Fined $26 million in 2024. The companies are now banned from deceptive telemarketing. The FTC continues to crack down on fraudulent tech practices previously targeting TurboTax, Avast and data brokers. Refunds will be sent via PayPal starting March 13, with recipients needing to redeem them within 30 days. Coming up after the break, my conversation with Gerald Bushelt, CISO et Acronis. We're discussing how threat research and intelligence matter to MSPs and the UK celebrates a record breaking Cyber First Girls competition. Stick around. Foreign threats are more sophisticated than ever. Passwords. They're outdated and can be cracked in a minute. Cybercriminals are intercepting SMS codes and bypassing authentication apps. While businesses invest in network security, they often overlook the front door, the login. Yubico believes the future is passwordless. Yubikeys offer unparalleled protection against phishing for individuals, SMBs and enterprises. They deliver a fast, frictionless experience that users love. Yubico is offering N2K followers a limited buy one get one offer. Visit yubico.com N2K to unlock this deal. That's Yubico. Say no to modern cyber threats Upgrade your security today.
