Loading summary
Transcript20 lines
- [00:02]
A
You're listening to the Cyberwire Network, powered by N2K.
- [00:12]
B
Heading to this year's Black Hat USA. The N2K CyberWire team will be on site recording from our podcast studio in the Spectre Ops Kennel Club. If you're interested in joining us for a conversation or learning more about what we're recording throughout the week, visit sponsor.thecyberwire.com for more information. And make sure you stop by the studio and meet the N2K CyberWire team. We'll see you there. What's the one thing in business that's spreading as fast as AI? AI risk. Every new tool your team signs up for. Every vendor that turns on AI features, every new integration. Each one is an opportunity for something to go wrong. And most security programs weren't built for AI's pace of growth. Enter Vanta. Vanta is the number one agentic trust platform used by over 16,000 fast moving companies like Ramp, Cursor and Harvey to ensure they're always audit ready. And now Vanta is helping companies like yours watch for the risks that show up between audits across your vendors, your AI tools and your whole environment. How the Vanta agent works like a 24.7grc engineer in the background, finding issues, drafting fixes and cutting vendor assessment time by up to 50%. Whether you're a fast growing startup or a global enterprise, Vanta is here to help you automate your security and compliance and earn and prove trust. Get started today@vanta.com cyber that's V-A-N T A.com cyber. Researchers track ransomware they say is getting gosh darn sophisticated. Zimbra patched a critical vulnerability affecting its classic web client. A sophisticated vishing campaign is targeting Microsoft 365 accounts. Gigawiper combines espionage capabilities with multiple destructive payloads. The EU sues member states over lax cybersecurity. The NSA revives Tao, a Puerto Rican agency exposes roughly a million Social Security numbers. A former ransomware negotiator heads to prison for assisting Black Hat. Our guest is Maxim Zavodchik, Senior Director of AI Security Research at Akamai with insights on the upcoming MCP specification and bad WI fi leaves a trophy up for grabs. It's Friday, july 10, 2026. I'm dave bittner and this is your cyberwire intel briefing. Thanks for joining us here today. Happy Friday. It is great as always to have you with us. We're a family show here. So in the interest of keeping things FCC friendly we are substituting the name of a new ransomware group with Gosh darn Researchers at Symantec say the latest version of the gosh darn ransomware marks another step forward in attacker sophistication. The malware is the newest evolution of the Hyadina ransomware family, which dates back to 2022 through earlier beast and Monster variants. In the observed attack, the operators used the remote access tool Anydesk, then deployed a malicious kernel driver called PoisonX that carried a legitimate Microsoft signature that allowed the attackers to disable endpoint security tools before installing credential theft utilities such as mimikats and Nierstroke Soft. After expanding their access and capturing administrator credentials, they encrypted victim systems and delivered a ransom demand. Researchers don't yet know how the attackers gained initial access or obtained the signed driver, but they say the campaign highlights how ransomware operators continue refining their techniques, particularly by improving their ability to evade security defenses. Zimbra is urging customers to immediately upgrade to fix a critical stored cross site scripting vulnerability affecting its classic web client. The flaw can be triggered through a specially crafted email, potentially allowing attackers to steal session data, mailbox contents and account settings when a message is opened. Although there's no evidence the vulnerability has been exploited in the wild, it was reported by Google's Threat Analysis Group, which has a strong track record of identifying threats used by advanced nation state actors. Zimbra has repeatedly been targeted by Russian linked hacking groups in recent years, with multiple cross site scripting vulnerabilities exploited to compromise governments and other high value organizations. The latest advisory reinforces the importance of promptly patching Internet facing collaboration platforms. Okta is warning organizations about a sophisticated vishing campaign targeting Microsoft 365 accounts active since April. The operation uses phone calls to convince employees they need to enroll a new Microsoft passkey then directs them to convincing fake Microsoft entra ID login pages hosted on attacker controlled domains. Rather than relying on automated credential theft, the attackers guide victims through the login process in real time, adapting the phishing pages to match each organization's multi factor authentication requirements. Researchers believe the goal is to trick users into approving an attacker controlled passkey, giving the threat actors persistent access to compromised accounts. The campaign has targeted organizations across multiple industries and highlights how attackers are increasingly exploiting users unfamiliarity with newer authentication technologies. Microsoft is warning about a sophisticated new malware platform called gigawiper, a modular backdoor that combines espionage capabilities with multiple destructive payloads. Active since at least October 2025, the GO based malware gives attackers persistent remote access while allowing them to trigger ransomware like encryption, overwrite files, wipe entire disks or even force systems to crash. Unlike traditional wipers that are built solely for destruction, gigawiper blends backdoor functionality with on demand sabotage tools, giving operators the flexibility to quietly gather information before launching a disruptive attack. Microsoft says the malware appears to share code with earlier destructive tools and may be linked to the developer behind Crucio Malware. The combination of stealth, persistence and destructive capabilities marks a notable evolution in wiper malware. The European Commission has preliminarily concluded that Meta may have violated the Digital Services act by using design features on Facebook and Instagram that encourage compulsive use. Regulators say Meta failed to adequately assess and mitigate the risks posed by features such as Infinite scroll, autoplay, push notifications and highly personalized recommendation systems, particularly for minors and other vulnerable users. The commission also found that existing safeguards, including screen time tools and parental controls are not effective enough to reduce excessive use. Officials say Meta should consider disabling some engagement driven features by default and introduce stronger protections. These findings are preliminary, and Meta will have an opportunity to respond before a final decision is made. If the commission ultimately confirms the violations, the company could face fines of up to 6% of its global annual revenue. Additionally, the European Commission is increasing pressure on several member states over delays in implementing key cybersecurity laws. Ireland, Spain, France and the Netherlands have been referred to the European Court of Justice for failing to fully adopt the European Union's NIS 2 directive, which strengthens cybersecurity requirements for critical sectors such as energy, healthcare, transportation and cloud services. The commission is also launching separate enforcement actions against Spain, France and Latvia, also over delays in implementing national enforcement measures for the Digital Operations Resilience act, or dora, which establishes cybersecurity standards for the financial sector. While most EU countries have now adopted the required legislation, the commission says timely implementation is essential to improving cyber resilience and ensuring consistent security and incident response across the bloc. The National Security Agency has revived the name Tailored Access Operations, or tao, for its elite Offensive Cyber Unit, reversing a restructuring that had folded the group into a broader Office of Computer Network Operations. The change is part of a wider reorganization intended to improve the agency's ability to respond to evolving cyber threats from countries including China and Russia. Former officials say restoring TAO also reunites developers and operators, a move expected to accelerate the creation and deployment of advanced cyber capabilities. TAO has long been associated with some of the NSA's most sophisticated hacking operations and became widely known through leaks involving the shadow brokers. The agency says reviving the name honors Tao's history or while positioning the organization for further cyber missions. Investigative reporting from Centro de Paradismo Investigativo and ProPublica found that Puerto Rico's Municipal Revenue Collection center inadvertently exposed the Social Security numbers of roughly 1 million people through its online property records system. Researchers discovered that attackers with basic technical knowledge could access sensitive data without authentication and alerted the agency in June. Although the vulnerability was quietly patched soon afterward, Krim continues to deny that any confidential information was at risk and has not notified affected individuals or Puerto Rico's Central Technology Agency. The incident is the latest in a series of cybersecurity failures affecting Puerto Rico's government. Security experts say many agencies have yet to fully implement cybersecurity standards required under 2024 law, leaving critical systems vulnerable to future attacks. A Florida cybersecurity professional who secretly worked with Black Cat or Alfie Ransomware Gang has been sentenced to nearly six years in federal prison. Prosecutors say Angelo Martino abused his role as a ransomware negotiator to provide the hackers with confidential information about victims negotiating strategies, helping maximize ransom payments. In exchange for a share of the proceeds, Authorities seized roughly $10 million in assets tied to Martino. He is the third former cybersecurity consultant sentenced in the case, underscoring the insider threat posed by trusted security professionals who betray their cl. Coming up after the break, my conversation with Maxim Zavodchik, senior director of AI Security research at Akamai. He has insights on the upcoming MCP specifications and bad WI fi leaves a trophy up for grabs. Stay with us.
- [13:20]
A
Foreign.
- [13:29]
B
Phishing attacks Faster, more convincing and harder for people to spot. And traditional security awareness and phishing training weren't designed for this level of attack. HOX Hunt helps security teams prepare employees for the attacks they face every day with personalized phishing training that adapts to each employee and reduces risky behavior over time. For IT and security leaders looking to strengthen their human layer of defense without adding more manual work. Visit hoxhunt.com cyberwire to learn more. That's h o-xhunt.com cyberwire. This episode is supported by Black Hat usa. If you follow the research, you know a lot of it breaks on Black Hat stages hundreds of peer reviewed briefings, more than 100 hands on trainings, and the largest business hall in Black Hat's history. Six days to learn the skills you'll need tomorrow, August 1st through the 6th, use code CYBERWIRE for $200 off your briefings pass@blackhat.com we'll see you in Vegas. Maxim Zavodchik is Senior Director of AI Security Research at Akamai. We recently got together to discuss insights on new security risks that can come from upcoming MCP specifications.
- [15:09]
C
MCP actually was invented last year and before MCP the models could reason could answer questions responsive from a knowledge base, but MCP actually allows connecting the agents to systems think about is like having arms now and legs to do operations. So MCP is the connecting protocol between the agents and the tools and functions they can do.
- [15:37]
B
So I know you and your colleagues at Akamai are tracking some changes to the MCP specification. What's caught your attention here?
- [15:48]
C
So that was interesting. I think that's kind of since its inception last year actually had a lot of different versions and variants and incremental ads, but this time it's different and that's how they said on the announcement page as well is that the upcoming specification actually represents the most significant architectural evolution of the protocol itself since since it began. What began as a local single user AI integration tool is transforming now to a platform capable and enterprise grade protocol to support businesses and build organizations exposing their services. AI native services using entity so what
- [16:33]
B
are some of the things that you think folks need to pay attention to here with these coming changes to be
- [16:40]
C
enterprise ready to make yourself suitable to the cloud architectures and proxy architectures. The protocol actually became stateless, meaning beforehand. Usually an agent will open a session that MCP protocol supports and then it's a long living session that you interact with the server back and forth by direction conversation. Currently it's a very stateless means that it will have request response interactions, client opens a connection, asks something and then close this connection. So it sounds technical, but actually that introduces also kind of a built in paradox here as well. So from one hand removing the session actually removes some of the threats related to the session such as session hijacking. But on the other hand actually it pushes the responsibility to manage those states to the application itself. The protocol is stateless now, but the application is not. And someone still has to remember which workflow is running, which tool invocation belongs to which user, which asynchronous task belongs to each tenant. MCP servers need to avoid predictable identifiers and poorly protected state objects. The state now belongs to the application means that the developers now own this state security.
- [18:05]
B
Now I understand there are also concerns about metadata injection.
- [18:10]
C
Yeah, besides moving to be stateless which is one of the biggest things, there's also More additions that actually introduce new attack surface. One of them is this metadata object which is not well defined, let's say like the object that the client can add to the request and also the server can add to the response, meaning that that's not structured or restricted by schema things which allows, you know, you can imagine creativity for developers, right? A new object that can flow on every request and response and actually introduce a new attack surface as well. For example, if some state information is being sent from the client to the server, but it's not properly signed or not properly validated at the server, that can create some problems there.
- [19:03]
B
And one of the other things that caught my eye, you all describe it as hit and run task abuse. What's going on there?
- [19:11]
C
So that's interesting. So as I mentioned, besides moving to statelessness now it also introduces MCP apps and tasks as first class citizen extensions. Something that was informally. Some implementations use this one, but now it's part of, kind of more formal part of the protocol. So tasks are those long running asynchronous workflows. Sounds great until attackers abuse them. One of the major threats as you mentioned, is the hit and run attack where client can open a connection, start a task and then disappear while the server still consuming heavy server resources for a longer time. Tasks are usually those long execution asynchronous workflows. And without proper timeouts on tasks and limiting the number of tasks, each client can start the service under serious resource exhaustion and denial of wallet threats.
- [20:13]
B
What are your recommendations for organizations to be best prepared for these changes?
- [20:20]
C
Yeah, that's a good question. I think it kind of goes to like whether those are new threads, evolving threads, right? So I think the MCP repeats web history of some threads feel familiar. IDOR or BOLA becomes workflow manipulation, protocol confusion, deterrence, stored. XSS returns now with the MCP apps. But agents change everything. I think agents need semantic context. They discover capabilities dynamically, intents become exposed and the coincidence becomes easier. On MCP servers, MCP tools become business actions, not just simple API calls and simple request response transactions. One tool, many systems, one request, entire workflows. MCP servers becomes brokers for sync capabilities, trust becomes centralized and the blast draguls grows as well. And security moves beyond just requests. Now you need to protect business workflows. And I think for the CISOs it's not just asking like one question. Do you support the new MCP protocol? But ask a better one, how is it implemented? Now it's all about the implementation. How do you validate state? How do you validate metadata? What guardrails exist? How do you prevent MCP apps and tasks abuse? The protocol is stronger by removing some of the vulnerability, vulnerable parts and, you know, related functionality. But the implementation matters. Trust boundaries, changed execution models, changed state management, changed security, moved up the stack. So the protocol isn't the risk. The implementation.
- [22:03]
B
That's Maxim Zavodchik from Akamai. As organizations grow, so does complexity. New applications are deployed, vendors are granted temporary access, and remote support tools are installed. Many of them never go away. In my recent conversation at RSAC 2026 with Rob Allen, Chief Product Officer at ThreatLocker, he explains how these forgotten tools create hidden pathways into enterprise environments and why attackers increasingly exploit what's already inside the network instead of trying to break through the perimeter. Learn how to reduce lingering access, shrink your attack surface, and implement zero trust more effectively by listening to the full conversation at explore.thecyberwire.com threatlocker.
- [23:10]
A
This episode is brought to you by Google Chrome. You think you know a browser, but Gemini and Chrome? That's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50 page restoration block. Or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it, ready to make anything online make sense. There's no place like Chrome. Check responses, set up, required compatibility and availability varies. 18 when you need to build up your team to handle the growing chaos at work, use Indeed Sponsored Jobs. It gives your job post the boost it needs to be seen and and helps reach people with the right skills, certifications, and more. Spend less time searching and more time actually interviewing candidates who check all your boxes. Listeners of this show will get a $75 sponsored job credit at indeed.com podcast that's indeed.com podcast. Terms and conditions apply. Need a hiring hero? This is a job for Indeed. Sponsored job.
- [24:12]
B
And finally, a story shared by the Register proves that if you carry a laptop with an antenna sticking out of it, people will assume you know what you're doing. Professional Red Teamer David Schloss was testing the physical security of a Fortune 500 company while its offices were under construction and the WI Fi was driving employees crazy. As he and his team wandered the building with conspicuous networking gear, no one questioned why they were there. Instead, they kept asking the same question. Are you here to fix the WI Fi? Schloss eventually wandered into the marketing department, opened a display case, removed a trophy worth hundreds of thousands of dollars, slipped it into his backpack, answered yes, when asked if he was fixing the WI fi and walked out. No one noticed the missing trophy for two and a half weeks. He finally returned it by placing it on the conference table before delivering his security briefing. The lesson here is simple. People tend to trust anyone who looks like they belong. Sometimes that's all an attacker needs. And that's the Cyberwire. For links to all of today's stories, check out our daily briefing@the cyberwire.com Be sure to check out this weekend's Research Saturday and my conversation with Jeff White, host of cyberhack and a BBC journalist. We're taking a closer look into his work into the Conti Ransomware gang. That's Research Saturday.
- [25:48]
A
Check it out and hello, Maria Vermazes here on Sunday's T Minus Space Cyber Briefing. We're diving into space cyber considerations for the fast approaching post Quantum World. That's in conversation with Eddie Zurvagon, CEO of Quantum Exchange. Hear more on Sunday on T Minus. Don't miss it.
- [26:06]
B
We'd love to know what you think of this podcast. Your feedback ensures we deliver the insights that keep you a step ahead in the rapidly changing world of cybersecurity. If you like our show, please share a rating and review in your favorite podcast app. Please also fill out the survey in the show notes or send an email to cyberwire. N2K's lead producers, Liz Stokes, were mixed by Trey Hester with original music and sound design by Elliot Peltzman. Our contributing host is Maria Vermazes. Our executive producer is Jennifer Ibin. Peter Kilpe is our publisher and I'm Dave Bittner. Thanks for listening. We'll see you back here next week.