CyberWire Daily - The guardrails go to court.

Episode Date: September 30, 2026

AI companies sign a voluntary accord aimed at addressing safety concerns. Cybercriminals abuse the CustomGPT feature as part of a ClickFix campaign. The FBI is urging members of ShinyHunters to come forward. A fraud campaign turns stolen credentials into job scams. Maria Varmazis joins us for her space cyber story. The WaterISAC confronts persistent weaknesses. AI gives SOC teams more time at the expense of training. Two U.S. Air Force members get federal prison time over a business email compromise scheme. Our guest is Dan Ohlemeier, Senior Solutions Architect for Ready1 at Semperis, discussing crisis orchestration. That is one big pile of technical debt. Remember to leave us a 5-star rating and review in your favorite podcast app. Miss an episode? Sign-up for our daily intelligence roundup, Daily Briefing, and you’ll never miss a beat. And be sure to follow CyberWire Daily on LinkedIn. CyberWire Guest On today’s Industry Voices we are joined by Dan Ohlemeier, Senior Solutions Architect for Ready1 at Semperis, discussing crisis orchestration: the unsung hero of successful cyber incident response. If you enjoyed this conversation, tune into the full interview here. Selected Reading Trump says top tech firms have signed accord to 'self-police' AI development (NPR) OpenAI Ignored Employees’ Warnings About Safely Testing A.I. Models (The New York Times) OpenAI Gets Sued Over the Hugging Face Hack (WIRED) Attackers Combine ChatGPT Feature Abuse With ClickFix to Deliver Trojan Malware (Infosecurity Magazine) ShinyHunters Defiant After FBI Calls on Members to Come Forward (SecurityWeek) From EDU Account Takeover to Job Scam Abuse: West African Fraud Actors Target Universities (Proofpoint) SpaceX's Starship returns early to Earth after reaching orbit for the first time (AP News)  WaterISAC reckons with range of threats after summer of cyberattacks (CyberScoop) AI Boosts SOC Analyst Capacity but Limits Skill Development (Infosecurity Magazine) US Air Force members given over 6 years in prison for cyber theft of more than $2 million (The Record) AI Finds a Way: The Jurassic Park Problem (Securonix) Share your feedback. What do you think about CyberWire Daily? Please take a few minutes to share your thoughts with us by completing our brief listener survey. Thank you for helping us continue to improve our show. Want to hear your company in the show? N2K CyberWire helps you reach the industry’s most influential leaders and operators, while building visibility, authority, and connectivity across the cybersecurity community. Learn more at sponsor.thecyberwire.com. The CyberWire is a production of N2K Networks, your source for strategic workforce intelligence. © N2K Networks, Inc. Learn more about your ad choices. Visit megaphone.fm/adchoices

You're listening to the Cyberwire Network, powered by N2K. When initial access vulnerabilities are cheap and automated, attackers can hit everything all the time. But at Black Hat, OpenAIs, Clint Gibler, and Spectorops's Robbie Winchester reminded me that we don't have to just fight AI with AI at machine speed. By taking a preventative mindset, defenders can mitigate attack pathways, so there's far less occurring. at machine speed in the first place. Listen to our full black hat conversation at explore. TheCyberwire.com slash specterops.

DLP sucks. Every CISO knows it. 20 years of rules and reg X built for another era. Jazz is the DLP that deeply understands your business. Melody, Jazz's agentic investigator, weighs every data movement against the data, systems, people in process, then provides analysts with the few incidents fully investigated. In 30 days, 2 million signals in, about 80 investigated incidents out.

No rules written. Jazz won the 2026 CrowdStrike AWS and NVIDIA startup accelerator from a nearly thousand applicants. See Melody inaction at jazz.security slash N2K. AI companies sign a voluntary accord aimed at addressing safety concerns. Cybercriminals abuse the custom GPT feature as part of a click-fix campaign. The FBI urges members of shiny hunters to come forward. A fraud campaign turns stolen credentials into job scams.

Maria Vermazas has the latest space cyber news. The water ISAC confronts persistent weaknesses. AI gives sock teams more time at the expense of training. Two U.S. Air Force members get federal prison time over a business email compromise scheme. Our guest is Dan Olimyre, Senior Solutions Architect for Ready One at Sempris discussing crisis orchestration. And that is one big pile of technical debt. It's Wednesday, September 30th, 2026. 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. President Trump and executives from several leading AI companies have signed a voluntary accord aimed at addressing safety concerns while keeping development moving. The agreement calls for robust internal controls, independent external audits, and board-level committees to review the findings while leaving open the possibility that those measures could eventually become law. Trump said the industry has demonstrated strong self-police.

and emphasized AI's potential for economic growth, while Anthropic CEO Dario Amadeh cautioned that the technology carries real risks. Critics argue voluntary oversight may be insufficient, and that slowing development could entrench the largest frontier labs. The administration is simultaneously pushing AI adoption across government, including through the new America.gov service, while framing AI development as a strategic competition with China. The announcement comes as OpenAI faces growing scrutiny over its security practices

after employees reportedly warned executives that advanced models weren't being adequately monitored during testing. Those concerns were brushed aside as the company pushed to meet release schedules, according to the New York Times. The models later escaped testing environments and autonomously targeted hugging face, and other organizations. OpenAI says it takes security seriously and has since paused some advanced training and withheld GPT 6.1 Astra over security concerns.

Now the Hugging Face incident is headed to court. Legal advocates for safe science and technology, or last, has sued OpenAI in California, alleging its agents violated the state's computer fraud law when they breached hugging face. The suit cites a new California AI law specifying that autonomous AI behavior isn't a defense against liability.

Last isn't seeking damages. It wants an injunction barring open AI from developing agents capable of autonomously hacking other organizations, potentially making the case and early test of legal responsibility for rogue AI agents. Cybercriminals are a lot of,

abusing ChatGPT's custom GPT feature as part of a click-fix campaign that Huntress says has infected at least 40 systems since September. Attackers created a custom GPT called Plus 5.6 that impersonated ChatGPT and was promoted through sponsored Google search results. Victims were told the service had limited availability and directed to a backup site displaying a fake cloud flare capture. The supposed verification instructed users to paste and execute a command, the hallmark of a click-fix attack. That ultimately delivered a malicious installer that used a legitimate, canon-signed application to side-load malicious code and install a remote-access

Trojan. The malware can steal data, capture microphone and camera feeds, and install additional payloads. OpenAI removed plus 5.6 after Huntress reported it, but researchers have already identified another custom GPT associated with the campaign. The FBI is urging members of shiny hunters to come forward following the arrest of suspected group leader Pepin van der Stop in the Netherlands. Dutch authorities arrested the 24-year-old on September 15th and say information found on his laptop also suggests he may have ordered two murders abroad. The FBI alleges Van der Stap helped Shiny Hunters compromise more than 140 organizations since 2025

and collect at least $70 million in extortion payments. The arrest follows Shiny Hunter's claimed breach of FBI jobs.gov The group insists its threats involving allegedly stolen FBI data were a publicity campaign rather than extortion, while separately warning other victims to keep paying to prevent data leaks. Security experts say Vanderstapp's arrest is unlikely to end Shiny Hunters, describing it as a decentralized operation whose remaining members could reorganize, rebrand, or join other groups. researchers at ProofPoint are tracking a fraud campaign targeting U.S. universities that turn stolen credentials into job scams. Attackers begin with emails warning that university accounts need verification

or risk deactivation, directing victims to forms hosted on legitimate services such as Google Forms, Wix, and Microsoft. The forms, steal credentials, and personal information, allowing attackers to hijack dot edu accounts and use their institutional credibility to advertise fake jobs. Victims who respond are sent fraudulent checks, typically around $1,000, and told to deposit them, keep part as wages, and use the rest to buy gift cards. The scammers then demand the card codes before the checks bounce. Researchers who engaged the fraudsters saw them escalate to alternative payment methods, harassment and even impersonating an FBI agent.

Proofpoint traced the operators it engaged to Nigeria and recommends MFA as a key defense. Maria Vermazes is host of the T-minus Space Cyber Podcast, and each Wednesday she brings us an update from her show. Today, she files this report. Thank you, Dave. The 14th time was the charm this week for SpaceX's super-heavy rocket, the starship, It hit a major and much-anticipated milestone on Monday when several years into its development

Starship has flown in orbital space for the first time. The 14th test of Starship did not hit every objective, but it did successfully deploy 26 Starlink version 3 satellites into low Earth orbit, making this flight a revenue-generating one for SpaceX. Now, to understand why the space industry has been eagerly awaiting milestones like this one with starship, we need to do some quick comparisons. The entire starship is built to be reusable for one thing. The current workhorse rocket for getting things into space, the SpaceX Falcon 9, has a reusable booster only. Additionally, the Falcon 9 is a much smaller vehicle compared to the starship, as the Falcon 9 is able to carry about 50,000 pounds of cargo to low Earth orbit in its 17-foot diameter

faring. Starship, on the other hand, has a nearly 30-foot diameter, And SpaceX says that it should be able to carry over 100 tons to orbital space. And that capacity difference is the key. Getting much bigger things into space all at once, instead of things being folded intricately or assembled slowly bit by bit, will be a paradigm shift for what kinds of things we can put into space. Imagine the International Space Station being assembled and, say, two large pre-built parts instead of bit by modular.

bit as it was for many years. Larger spacecraft means more power generation, which unlocks more complex capabilities, including more carrying capacity for crew, instrumentation, and other components. And space agencies like NASA also hope that this will lead to more complex infrastructure on orbit, one day on the moon, and then one day beyond. After all, both SpaceX's starship and competitor Blue Origins Blue Moon will be tested in NASA's Artemis 3 mission later next year. That will be in preparation for a human moon landing using either the Blue Moon or Starship for the Artemis 4 mission, which currently is planned for early 2028. But as always, time will tell.

For the CyberWire Daily, I'm Maria Vermazes from T-MIS space cyber briefing. Back to you, Dave. Be sure to check out the T-minus Space Cyber Podcast, wherever you get your favorite shows. The U.S. Water Sector is confronting persistent cybersecurity weaknesses after a summer of attacks highlighted the risks posed by aging operational technology. Water ISAC executive director Tom Dobbins says, Internet-exposed OT and programmable logic controllers remain major entry points

because much of the equipment predates modern cyber threats but remains operational, reducing incentives to replace it. Other risks include poorly secured connections maintained by third-party integrators, fishing attacks against employees, and weak cyber hygiene at smaller utilities, where basic measures such as password changes and multi-factor authentication, can be challenging to maintain. Water ISAC is responding by partnering with SciWARE to accelerate threat intelligence sharing, including across critical infrastructure sectors.

Dobbins said utilities face threats associated with Iran, China, and Russia, while financial and technological constraints continue to complicate efforts to modernize defenses. AI is giving security operations teams more time for higher value work, but it may also be disrupting the traditional path for developing SOC analysts. That's according to a swim lane survey. of 500 U.S. and U.K. security professionals. Nearly half said AI has increased their capacity, while 35% reported having more time for complex investigations and strategic work.

The trade-off is training. 24% said AI has limited their ability to develop skills, as automation increasingly handles the routine tasks through which junior analysts traditionally learn the fundamentals. Nearly half expect AI to make entering the profession more difficult, including 37% who anticipate higher requirements for entry-level positions. Swim Lane argues organizations will need more deliberate training and career paths while maintaining human oversight of AI decisions. 41% expect new roles focused on AI oversight, validation, and orchestration.

Two U.S. Air Force members have received received federal prison sentences for running a business email compromise scheme that stole millions of dollars. Shizoke Timothy Odomegu and Harifat Maggi conducted the operation for more than two years while stationed at Dover Air Force Base in Delaware. Prosecutors say the pair used fishing to steal employee email credentials, then monitored compromised accounts for payment discussions. They inserted themselves into email conversations. and supplied fraudulent wiring instructions, including schemes that diverted payments of over $2.5 million.

They also stole financial account and payment card information for their own use or sale to other hackers. Both pleaded guilty to wire fraud, identity theft, and access device fraud. Odomegu received more than nine years in prison, while Magawi was sentenced to six and a half years, with both also ordered to pay, Destitution.

Coming up after the break, my conversation with Dan Olmeyer from Sempris, we're discussing crisis orchestration. And that is one big pile of technical debt. Stay with us. Every time your team deploys a new cloud workload or AI agent, another identity gets permitted access to your critical systems. Legacy tools were built to manage human employees, leaving modern machine and AI, access largely unmanaged. That's where IDERA by Palo Alto Networks comes in. Human, machine,

AI, one identity platform for all. Idira replaces permanent permissions with dynamic access, so you can lock down every identity without slowing down your business. Secure every identity with IDERA by Palo Alto Networks. Visit Palo Alto Networks slash IDERA. Again, that's Palo Alto. Networks.com slash IDIRA. Maybe that's an urgent email from your CEO, or maybe it's a deep fake targeting your business. Dopple is the AI-Native social engineering defense platform

fighting back against impersonation and manipulation. As attackers use AI to make their tactics more sophisticated, Dopple uses it to fight back, automatically dismantling cross-channel attacks, building team resilience, and providing agentic email protection. Dopple, outpacing what's next in social engineering. Learn more at doppel.com. That's do p-p-p-el.com. Dan Olemire is Senior Solutions Architect for Ready One at Sempris. In today's sponsored industry voices conversation, we discuss Crisis Orchestration.

One of the things I've observed in my experience going through the incident response, investigation and recovery work that happens when cyber level crisis events happen, we often focus a lot on what we need to do, but we don't plan ahead and think about how we're going to do those things. Something to consider would be, you know, how am I going to contact that site engineer in the event that my network is down? Am I going to have access to my address book? Will I even be able to know what they're,

phone number is or, you know, if I need to share documents with them, how am I doing that if our file sharing site is offline? So quite often, organizations are figuring these solutions out on the fly in the midst of their crisis response. And the big challenge with that is you're just, you're losing time and expending resources on some things that would have been probably better, better set up in advance. Why do you suppose these gaps are so persistent? that you see them over and over again? I think a lot of it has to just do with, you know, we're at along the journey of resiliency, I would say.

And some of it's just the nature of Infosec. For example, if I have the resources to do 70 things, but I know I have 100 things to do, I'm going to have to make some educated decisions on things that I kind of push down the line a bit, right? and prevention usually trumps any sort of a resiliency type solution because we know we've got to be preventing things. We know we've got to be continuing to chase some of the technological advancements

and changes. And finding the energy and time to invest in those worst-case scenarios can just be a little tougher to do sometimes. I know a lot of folks will invest time and energy in doing tabletop exercises. Where do they come up short when it comes? comes to that particular process? That's a great question.

So I applaud any organization that spends any amount of time doing any sort of theoretical exercise and simulation work. That's great. The more formal, great, even informal stuff's effective. What I do want to say, though, is some observations I observe over time is we see a bit of a focus on testing the infosec team and a bit of a focus on that executive level response. but sometimes we miss some key players like bringing in your crisis management focals

or perhaps an application engineer. I know that when I've had to go chase down hosts before because I'm investigating some interesting activity, it's the application engineer who's going to be my connection into that host. Well, if they don't know appropriate evidence collection procedures or what the break class recovery process is and the first time they're learning it is in the midst of a crisis that can have

some unintended consequences and slow down your ability to effectively recover. And it sounds like a good tabletop exercise would reveal what some of those dependencies are between the various stakeholders in the organization. Absolutely. I do think the more practice that occurs, you know, when organizations start getting into a, you know, a quarterly cadence of table top exercises and start having more operational-focused ones, that will uncover those, those gaps. right? You know, maybe you'll do your post-incident or your

post-table-tap activity and review and you'll go, hey, we need this new crisis management focal that got hired a month or so ago that this one business unit stood up. They weren't actually included in this conversation. Now you make that connection and then you can start to get that practice

in before you're going live with your response. What about the crisis orchestration technology itself? What kind of tools are we talking about here and what role do they play? That's a really good question too. There's a lot of technology that you can bring to bear. So, you know, thinking about what you're going to need to do in times of crisis,

you're going to need to manage the response, you're going to need to document your response, you're going to need to communicate your response, and so then what technologies do you need? Communication, you're going to need internal and external, right? So, like, if I'm an Infosec team or a crisis management team, you know, I need to orchestrate my response with all my stakeholders across the organization, you know, for rebooting systems and collecting evidence and deploying technologies, all the things that we have to do when we're

responding. And then mass notification too, right? Like, how am I communicating with all of my workforce, you know, because it's critically important to keep them informed. But many organizations are thousands of people strong. So having a tech. that facilitates, that's great. When it comes to documentation, a lot of organizations kind of devolve to using Microsoft Excel. It's a classic for a reason. But it doesn't have really good version control. It's not built with security in mind.

And oftentimes, organizations haven't necessarily thought about what to document. So there might be gaps in timestamp records that you're collecting or gaps in the types of logs that you might want to be able to memorialize in your report. So, you know, thinking about systems where you can collect and share this information, often the connecting tissue with all these systems would be they need to be out of band your network because, you know, I think for some of the most serious crises where there's a total network loss, quite often organizations are going through the process of recovering their identity systems first just so they can access their insurance policy because it's on a file share that's currently inaccessible or access that incident response plans.

So they're going off of a paper copy they might have around or they're going off a memory. So having these Ataband tools staged and having people practice with them is really important. Yeah, it's a great point. I mean, I guess a lesson here could be, don't underestimate the value of having binders full of printed out documents

sitting on some important person's shelf. Yeah, absolutely. You know, it's funny. We've done such a phenomenal, job digitizing all of the things we do for work, right? So now it is just as important, if not more so, to have those old school, like have a file cabinet that you keep copies up to date in and maybe have a couple fax machines in a couple locations because you can send stuff pretty securely over fax.

You know, so it's never a bad idea to roll out looking at old proven methods for keeping information. And getting back to the tabletop exercises, is it. it kind of human nature for people who are busy, you know, having the tabletop exercise may not be all of the stakeholders' primary function in the organization. So is, do you, are you sympathetic to that some things tend to get hand-waved away? I'm hugely sympathetic to any organization that's tackling really anything in information security. Because, I mean, you're, you're battling the torrent of the internet right at your doorstep.

So, yeah, you know, there's some things organizations can do to kind of help themselves be a little more successful. I think a lot of it's a state of mind to how you approach these exercises. You know, some organizations, there's a concern, right? And like in business and in life, we're very averse to failure. You know, it's all about striving for success. If you can bring a failure mentality to your tabletop exercises where you explicitly set the goal is to fail, and you can really embrace that concept, then it's a lot easier to uncover these gaps

in your policies and your procedures and technologies that you have that are going to orchestrate all this and all the things that go into crisis response. So there's really a culture component to this as well, giving people the encouragement that failure leads to learning. Absolutely. It does really help significantly when you're trying to exercise

a really effective crisis management program, kind of overseeing your incident response and your recovery efforts that, you know, if you've got a really good culture, that kind of takes a team effort and is really interested in the learning and the practicing of the things, right? You know, because we're always striving for mastery. I consider myself a perpetual learner in the space, right? Because there's just so much. Help me understand the importance of documentation here. I mean, you're in the middle of a serious incident, the heat of battle, if you will. I can understand how people might say to themselves, you know, we're going to, we'll document this when we get through all of this. But you say no,

it's important to keep that documentation running in the midst of things. It is so critically important. The moment your crisis level event starts happening should be the moment that you start collecting and documenting your effort. Because it's infinitesimally easier to note things as you're going through the process than it is to retroactively recreate that. I'm documenting a timeline of events. If I take note of all the things I'm doing and write down the time that I do those things, it goes a long way. One of the things that I've observed with organizations is, you know, when it comes to the end of your crisis, you know, we have stakeholders that are going to want to know some things. Maybe customers have agreements to get

some reports on what happened, because, you know, especially in technology, right, if we offer a service or something, you know, our customers are going to want a good idea that that saves. So, so being able to build a really robust timeline of events that tracks the timestamp and all the actions that happen across your organization from the moment this thing happened to when you're, you know, back online is really important. That gets appended to a lot of reports. And that's something regular readers are going to want to look at regarding, like, insurance claims. If there are gaps in some of your records keeping that can have negative impacts on your claim payout. So there's a tremendous amount of reasons to be really good at documentation.

And the best thing you can do is to just prepare ahead of time. You know, talk to your legal counsel experts and talk to your insurance providers and talk to the business stakeholders and some of your customer stakeholders on some of the key information that they're going to want documented. in a crisis and then just make sure that you start collecting that right from the get-go. Have that built into your response plan that when you start doing all these tasks, maybe even put a sub-note and make sure you document it here. That's Dan Olomeyer, Senior Solutions Architect for Ready One at Sempris. Hey, everybody, Dave here.

I want to let you know about a special gathering hosted by Zimperium at the Spy Museum in Washington, D.C. This invitation-only event will bring together federal cybersecurity and technology leaders to discuss some of the most pressing challenges facing government today, including mobile security, mission resilience, and the evolving threat landscape. I'm always grateful for opportunities to spend time with smart people doing important work, and I'm excited to be part of these conversations. I hope to see some familiar faces there. If you're interested in attending, you can request an invitation.

You'll find more information in our show notes. Our thanks to Zimperium for sponsoring this event. We'll see you there. And now a word from our sponsor, SpectreOps. Today, AI is rapidly adding non-human and agentic identities to modern enterprise environments, creating new trust relationships and attack paths. Bloodhound Enterprise helps defenders map attack paths across AWS and hybrid environments as one connected graph, identify the choke points that matter most, and bring trusted attack path

intelligence into approved AI workflows with Bloodhound Hunter. See how SpectorOps helps team secure the AI-driven identity era at Spectorops.io. And finally, Securonics researcher Aaron Beardsley argues that AI's biggest security problem is increasingly familiar. Capability is advancing faster than containment. The clearest example is OpenAI's cyber evaluation that escaped its intended boundaries and reached hugging face, where agents reportedly executed code on dozens of servers and gained route access to one, life, or at least AI, found away. Beardsley warns the problem will grow as agents help develop future systems, while simultaneously gaining access to enterprise tools, data, APIs, and workflows.

A sandbox is only as good as its overlooked exits, and agents are quite willing to keep testing the fences. His prescription is less cinematic, stronger egress controls, meaningful independent evaluations, detailed logging, tightly constrained permissions, and human approval for consequential actions. Enterprises shouldn't work. wait for frontier labs to slow down. Once AI agents can act on production systems,

governance needs to be built before someone discovers the raptors can open doors. I'm simply saying that life finds a way. 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. The 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 and the show notes 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 Kilpy is our publisher. and I Dave Bittner. Thanks for listening. We'll see you back here tomorrow.