CyberWire Daily - Cybercrime finds its sea legs.

Starting point is 00:00:00 You're listening to the Cyberwire Network, powered by N2K. So what happens when an AI agent isn't malicious, but still does something it shouldn't? I recently sat down with Cal Al-Dubabe, principal technologist at Rubrik, to talk about why agentic AI is challenging the way security teams think about detection, permissions, and recovery. If your organization is deploying AI agents, this conversation will help you think differently about where the risks are and how to prepare when things go wrong. Listen to our full conversation at explore. thecyberwire.com slash rubric. 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

Starting point is 00:01:05 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, Hercer, 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. The Vanta agent works like a 24-7 GRC engineer in the back. background, finding issues, drafting fixes for you, 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 at vanta.com slash cyber. That's v-a-ta.com

Starting point is 00:01:59 slash cyber. Officials investigate suspected cyber attacks on U.S.-bound oil tankers. Iranian operators deploy chosen brick surveillance malware. Ukraine cracks down on scam call centers. Researchers uncover two TP-link camera zero days. Maria Vermazas looks at weapons in space. Centerpoint energy reports a data breach. Spain records its first breach caused by an autonomous AI agent.

Starting point is 00:02:41 Phantom Raven targets developers. through malicious NPM packages. Illicit casinos provide cover for cybercrime. A New York health care provider exposes patient data. Our guest is Chad Tunberg, Sissau at Ubiko, on how real crypto agility still needs a hardware route of trust. And hackers do a little flock picking. It's Wednesday, September 16, 2026.

Starting point is 00:03:16 I'm Dave Bittner, and this is your Cyberwire Intel Briefing. Thanks for joining us here today. It's great as always to have you with us. The Coast Guard and FBI are investigating suspected cyber attacks against two oil tankers bound for the United States last month. One vessel was reportedly compromised while transiting the Strait of Gibraltar, losing communications for more than 30 hours.

Starting point is 00:03:56 On August 21st, a specialized Coast Guard and FBI team boarded the foreign flag tanker in the Gulf of Mexico to assess and secure its operational and information technology systems. A second tanker, also reportedly targeted, was boarded three days later for a similar assessment. Authorities say neither incident caused operational disruption, vessel instability, danger to cruise, or environmental damage. Investigators are examining whether Iran or another foreign actor may have been responsible, potentially in connection with ongoing tensions between Iran and the United States.

Starting point is 00:04:38 Cybersecurity and intelligence agencies in the U.S., U.S., U.K., and Netherlands are warning about chosen brick, a Windows malware family used by Iranian state actors to surveil dissidents, activists, and journalists worldwide. Active since at least 2025, the malware can steal emails and messages, capture screenshots, record microphone audio, track victims' activities, and even wipe data. Operators typically approach targets through WhatsApp or Telegram, posing as acquaintances or technical support before delivering malicious files disguised as utilities or medical documents. If corporate security blocks the attack, they may try moving victims to personal devices.

Starting point is 00:05:24 Once installed, Chosen Brick establishes persistence, attempts to weaken Microsoft Defender protections and uses telegram and cloud storage for command and control data exfiltration. Authorities say the campaign supports Iranian state-sponsored repression and harassment of perceived opponents. Ukraine's parliament has approved tougher penalties, targeting fraudulent call centers and electronic fraud, following a corruption investigation involving alleged protection payments to prosecutors. The legislation makes using electronic communications to commit fraud a separate crime and criminalizes organizing, operating, working for, or recruiting for scam call centers.

Starting point is 00:06:11 Convictions could bring prison sentences of seven to 12 years. The bill, first introduced in 2023, gained momentum after anti-corruption investigations allege that officials accepted bribes beginning in 2025 to shield suicide. scam operations targeting victims in Ukraine and abroad. Five people have been named as suspects. Prosecutor General Ruslan Kravchenko, who denied protecting scam centers and has not been charged, resigned and was subsequently dismissed. The legislation now awaits President Zelensky's signature. For decades, the idea of weapons in orbit has lived somewhere between Cold War planning and science fiction, Maria Vermazas joins us with what we know about America's newly acknowledged arsenal in the final frontier.

Starting point is 00:07:08 Thank you, Dave. On Monday, the Pentagon publicly acknowledged that the U.S. Space Force not only has weaponry in space, but that the weaponry has already been deployed. Secretary of the U.S. Air Force Tri-Mank said this during a speech at the annual Airspace and Cyber Conference, and I quote, Today we continue to ensure we remain ready to meet the challenges of evolving threats wherever they exist. This is why the United States now has on-orbit space control weapons capable of defending the joint force against hostile adversary actions. The exact nature and number of the weaponry is not known, but this acknowledgement is extremely notable. It has been widely believed for some time now that the United States has some kind of space. weaponry to counter threats on the ground or on orbit. But this statement was essentially saying the quiet part out loud.

Starting point is 00:08:03 The first time a U.S. official has publicly confirmed the existence of such space weapons. And while the phrase space weapon might spring to mind something kinetic to potentially destroy, for example, an adversarial satellite, physically destroying a satellite could create a cloud of space debris and disable whole swaths of orbital space from use for decades. essentially causing far more problems than it would solve. Instead, it is far more likely that space weaponry or counter space in the lingo would either use cyber attacks, electromagnetic spoofing or jamming,

Starting point is 00:08:38 or laser dazzling, and yes, that is the term, to jam signals or up or down links, or otherwise non-kinetically disable a target satellite. For the CyberWire Daily, I'm Maria Vermazes from T-Minus, space cyber briefing. Back to you, Dave. Be sure to check out the T-minus Space Cyber Briefing. You can find that wherever you get your podcasts.

Starting point is 00:09:02 Researchers at Opswatt have disclosed two zero-day vulnerabilities in TP-Links TapoC-200 security camera, commonly used in homes and small offices. The more serious flaw could let an attacker with network access bypass authentication and gain an administrative session, potentially exposing live video and stored recording. A second vulnerability could allow an unauthenticated attacker on the network to crash the camera's HTTPS service, causing a denial of service. TPLink patched both flaws in firmware released August 18th. OpsWod is also working with TPLink on a third undisclosed vulnerability. It rates as critical.

Starting point is 00:09:48 Researchers say that flaw could potentially allow full compromise of the camera and turn the device into a foothold for, further activity on the victim's network. CenterPoint Energy reported a data breach affecting some of its 7 million customers across four states. A threat actor claims an inadequately protected company API allowed the theft of 7.5 million records containing personal, account, billing, and potentially sensitive information, including driver's license data and partial social security numbers. CenterPoint hasn't confirmed that figure.

Starting point is 00:10:25 the utility has brought in outside cybersecurity experts and notified authorities. Electric and gas services remain unaffected and impacted customers will be notified as required. Spain's Data Protection Agency has reported what it says as the country's first personal data breach caused by an autonomous AI agent. According to AEPD President Francisco Perez Bez, an individual deployed an agent powered by an unnamed large language model against an organization. The agent scanned files, probed the target for vulnerabilities, and chained together multiple attack stages to gain read and write access to files containing personal data and invoices.

Starting point is 00:11:13 Perez-Bez said the incident demonstrates that AI-supported attacks are no longer theoretical and warned that their speed could challenge traditional defenses. He called for human oversight backed by automated detection, containment, and response capabilities. The case comes amid broader concerns about autonomous agents performing unintended or unauthorized actions against third-party systems. Crowdstrike has identified Phantom Raven, a JavaScript information stealer distributed through malicious NPM packages by a self-described bug bounty hunter. Researchers assess with high confidence that the malware was likely generated using a large language model, citing statistical token patterns, verbose comments, placeholder code, and unusual design choices. Phantom Raven uses typo-squatted packages and remote dependencies to execute malicious pre-install scripts.

Starting point is 00:12:15 Once running, it collects system details, user information, and CICD environment variables, that could expose credentials, then exfiltrates the data over HTTP. Crowdstrike says the operator appears to use compromised systems to identify potential bug bounty opportunities rather than sell stolen logs. The researchers say the campaign illustrates how AI-generated tooling can lower technical barriers and accelerate malware development even for relatively unsophisticated operators. Researchers at InfoBlocks say the sprawling ecosystem of illicit online casinos is providing cover for three distinct forms of cybercrime.

Starting point is 00:13:01 They track more than 1.7 million Chinese language gambling domains that facilitate illegal betting and money laundering, including activity tied to transnational organized crime. A second category, dubbed Scambling, uses fake or rigged casino sites to look at least. lure victims into depositing money they ultimately can't withdraw. The third category has little to do with gambling at all. China-aligned threat actors are disguising malware, command and control infrastructure as low-quality casino and adult websites. Info blocks says groups using the Peck-Burdy framework have employed the technique since

Starting point is 00:13:42 2003 in espionage campaigns against corporate and government targets across Asia. The sites can look nearly identical, making their very different purposes difficult to distinguish from appearances alone. New York health care provider Premier Medical Group is notifying more than 282,000 patients after attackers stole personal and medical information during a June cyber attack. PMG says attackers accessed files on June 14th containing names, contact details, birth dates, treatment, and diagnostic information. medications, insurance data, and other patient records. The company has not disclosed how attackers gained access or identified who was responsible. No known ransomware or extortion group has publicly claimed the incident. Coming up after the break, Chad Tunberg from Ubiko discusses how real crypto agility still needs a hardware route of trust,

Starting point is 00:14:53 and hackers do a little flock picking. Social engineering attacks look trustworthy, a routine request, an internal email, a familiar face on a call. But Dopple sees through the disguise. Their AI-native platform detects and disrupts attacks across every channel, trains employees to recognize deepfakes and deception, and investigates every fish to take down the campaign behind it. They fight relentlessly to protect your business, brand, and people. Dopple, outpacing what's next in social engineering.

Starting point is 00:15:40 Learn more at doppel.com. That's do p-p-p-el.com. Chad Tunberg is Sissau at Ubiko. We recently got together to discuss how real crypto agility still needs a hardware route of trust. A hardware to route of trust is important because the system security model necessitates that the secrets that are used for core cryptographic operations

Starting point is 00:16:21 are protected in a way. way that minimizes its exposure to an attacker. So if you store secrets in an operating system or a piece of software, there's a lot more exposure or attack surface for the attacker to compromise that information. And is part of this, the sort of fundamental immutability of hardware itself? It is. We want to be careful with absolutes in the security industry. Fair enough.

Starting point is 00:16:52 That is intense, yep. Well, as we are on this sort of collision course with post-quantum cryptography, how does this come into play? Why does this become particularly important? Well, for a couple of reasons. So the executive order that was released this year that provided direction on post-quantum readiness is really starting to focus on the software side where the non-hardware route of trust And I think that's for a variety of reasons. One is that the information that's traversing the Internet or networks is really the information that's at highest risk.

Starting point is 00:17:36 And we can accomplish an acceptable amount of protection for that information in the near term. Eventually, we'll need to replace quite literally things like motherboards that have TPNs and our smart cards and our passports and all these other things. physical things that provide us the hardware rate of trust. But that's a little longer timeline than I think that we have today, as we consider a post-quantum threat. How should organizations be prioritizing their efforts then in terms of being prepared for this? Well, hopefully most organizations already have a business continuity and disaster recovery function that is considered some of their most business critical applications and processes. We can leverage some of that work to create an inventory of systems or assets that are critical

Starting point is 00:18:34 to keep confidential or secret and use that list of assets as the way to prioritize our effort as we prepare for a post-quantum future. Practically speaking, what that means is that the underlying cryptography that's protecting information using encryption or authentication will need to be updated, starting usually with software and then potentially with hardware, depending on how the organization is storing information, storing secrets, and providing that hardware route of trust. What about the influence of AI? You know, as we've got all of these non-human identities, coming online, what factor does that play in all of this? Yeah, the EA agents have really created a scale problem, especially with non-human identities.

Starting point is 00:19:35 Authentication relies on, by and large, relies on a signature process, which is kind of a different form of cryptography than the way in which we encrypt information to protect it from, crying eyes. There's a lot of work being done still on post-quantum signature algorithms. And we expect that the innovation for authentication in general, including non-human identities, is going to continue to evolve, maybe up until the 2030, 2013, 2021 dates that are talked about quite a bit. So what's your advice for organizations that are looking to get on top of this? From the hardware point of view, what are your words of wisdom here? Because hardware is going to have a bit of a longer timeline than updating software,

Starting point is 00:20:27 it makes sense that start thinking about that as soon as possible, whether it's phytal authenticators that store pass keys or TPMs that are sitting on motherboards or even badging systems that you're using for your facilities access. developing a logistics plan in order to look at an upgrade path or a replacement path for that hardware really necessitates a certain amount of planning today. So, Chad, am I going to have to replace my ubiquies? You will. In a post-quantum era, one of the challenges, which is not unique to a ubiquit,

Starting point is 00:21:09 which is why we need to replace motherboards, is that these post-quantum algorithms that are going to be used to ensure that our information is safe and that our authentication flows are safe, require a lot more processing power. And then quite literally the chips and the amount of memory that we gave those chips to use is just not enough. It's not enough processing power,

Starting point is 00:21:33 not enough storage power. And so we'll see a need to replace a majority of it over time. All right. Well, I think I have everything I need for our story here. Is there anything I missed, anything I haven't asked you that you think it's important to share? I think one of the salient points is that, you know, CISOs and security professionals, general priority is really based on what is happening in the moment.

Starting point is 00:22:04 We are all dealing with incidents and emergencies and urgencies across an organization. And the messaging around post-quantam is not dissimilar to how we thought about Y2K is it's going to take a fair amount of planning, coordination, testing in order to get to a point where we're confident that we'll have a resilient and robust environment to continue to protect the things that we care a lot about as both individuals and organizations. Not all of us were around as part of Y2K. Some of us were lucky enough to be in the workforce after having to deal with that. But there's a lot of parallels in the amount of planning and execution that needs to happen in parallel. And I think it's a good way to see a pattern that existed previously that we can crib from a bit.

Starting point is 00:22:58 It's an interesting comparison. One of the things I wonder about is obviously Y2K had a specific deadline. We knew when that moment was going to occur. And yet I hear a lot of folks wondering if with, With quantum, could we be in for some kind of a Sputnik moment where either us or one of our adversaries makes some sort of advance that was unexpected and moves the timeline? That is the concern. You know, thinking about timeline, unfortunately the timelines now or the timeline was quite some time ago, really because of the threat of harvest now and decrypt later, right? The information we care about right now is already at risk and at risk in a way that is going to probably lead to a specific moment.

Starting point is 00:23:50 The goal I feel is to really get us as far down that path as possible and then to try to be conservative on the timeline, right? Like, is it actually 2030? Is it actually 2031? No, probably not, right? But if we plan towards those specific dates and timelines, when we have that moment, hopefully as an industry or as an organization, will be less surprised by that moment and feel like we are prepared and maybe just need to accelerate. That's a really great parallel.

Starting point is 00:24:28 Think about, you know, how this, how this, how this. breakthrough, this innovation is going to come through. I can't think of many things that have happened that are on the innovation fraud, which post-quantum certainly is, where we knew the date that everything was going to come together and be a reality. That's Chad Tunberg from Ubiko. And finally, a group of hackers decided that simply tearing down a flock safety license plate camera wasn't enough. They took it apart, copied much of its storage, recovered an encryption

Starting point is 00:25:21 key, and handed the results to wired and 404 media for a look under the hood. What they found was a remarkably busy little roadside computer. Across roughly 21 days of recovered logs, the camera photographed about 50,000 vehicles and generated 1.6 million images. A typical passing car prompted about 28 shots, while some got more than 100. The camera's software detects vehicles, license plates, bicycles, and notably people, although researchers found no evidence that Flock's software was performing facial recognition. The system also occasionally got creative. Its license plate detector mistook bumper stickers, dealership frames, and even an American flag patch on a motorcycle for license plates. The findings raise questions.

Starting point is 00:26:14 about flocks on-device encryption. Much of the sensitive storage remained inaccessible, but the hackers found an encryption key elsewhere on the device that unlocked stored images and video. Flock says it hasn't received the findings through its vulnerability disclosure program and doesn't have enough information to assess the claims. And surveillance technology apparently has mundane problems too.

Starting point is 00:26:41 Logs contained more than 27,000 no-space, left on device errors, plus crashes and reboots. A watchdog process, nevertheless, kept checking in every couple of minutes with, who's a good boy? Even the surveillance state, it seems, appreciates positive reinforcement. And that's the Cyberwire. For links to all of today's stories, check out our daily briefing at thecyberwire.com. We'd love to know what you think of this podcast. Your feedback ensures we deliver the insights that keep you a state. 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

Starting point is 00:27:37 or send an email to Cyberwire at N2K.com. N2K's lead producer is Liz Stokes. We're mixed by Trey Hester with original music and sound design by Elliot Peltzman. Our contributing host is Maria Vermazas. Our executive producer is Jennifer Ibin. Peter Kilby is our publisher, and I'm Dave Bitner. Thanks for listening. We'll see you back here tomorrow.

CyberWire Daily - Cybercrime finds its sea legs.

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.