The short version: in 2012, Chinese state-sponsored operators hit PetroVietnam, the Vietnam News Agency, and Vietnam Post in a coordinated campaign timed to the South China Sea territorial pressure Beijing was running simultaneously. Three targets, three different intelligence functions, one unified strategic moment. Fine. That's the event. Here's what it actually tells you about where this goes. The target selection is the thing worth sitting with. PetroVietnam gives you drilling data, concession boundaries, what Vietnam knows about what's under the water Beijing is disputing. The Vietnam News Agency gives you visibility into the domestic narrative, what the government is telling its own people, where the soft spots in public messaging are, and potentially a position to shape what gets reported if you ever get deep enough. Vietnam Post is the one that gets underestimated. Postal and communications infrastructure is a long-game play. You're not there for the letters. You're there for the metadata, the routing, the correspondence patterns of officials and journalists and business contacts. That's counterintelligence value. That's map-building for future operations. What you're looking at in 2012 is a government that had already figured out that territorial disputes don't happen in one domain. The drilling rig standoff, the diplomatic pressure, the cyber operations: these weren't running in parallel by coincidence. Someone was coordinating across lanes. That level of multi-domain synchronization is not improvised. It's institutional. It means there are people inside the Chinese government whose job is to make sure the hackers are reading the same calendar as the foreign ministry. That matters enormously for how you project the trajectory, because institutional capacity compounds. It gets better at this. The 2012 campaign was, by the standards of what came after, relatively blunt. Named targets, recognizable infrastructure, attribution that intelligence services could make with reasonable confidence based on technical indicators and timing. The operators learned from that exposure. They always do. By the middle of the decade, the same broad operational concept, gather energy intelligence, monitor and potentially influence media, map communications infrastructure, was being executed with considerably more patience and considerably less visibility. Longer dwell times. More care about blending traffic. Broader use of third-country infrastructure to muddy attribution. The Vietnamese were not the only ones watching this evolve. So were the Filipinos, the Malaysians, the Indonesians. Everyone with an overlapping claim and an energy sector watched the 2012-era operations and either upgraded their defenses or didn't, and the ones who didn't paid for it later. The trajectory for targeting Vietnam specifically is not hard to read. PetroVietnam has continued expanding offshore operations, including in blocks that China disputes. Every new concession, every new exploration deal Vietnam signs with a foreign energy company, is an intelligence collection priority for Beijing. The cyber operations follow the money and the maps. When Vietnam signed agreements with foreign operators to drill in contested waters, the intrusion attempts tracked those agreements. This is not speculation. It's the pattern that played out repeatedly across the region. The Vietnam News Agency targeting has implications that extend well past 2012 and well past Vietnam. State media in authoritarian-adjacent or single-party systems often serves as a formal and informal channel between the party and the public. If you can read the internal deliberations of a state news agency, you get visibility into what the government is planning to say before it says it. That's genuine strategic value. And if you can get persistent access, you eventually get to ask whether you can influence what gets drafted, not just read it. That second step is the one that intelligence services outside China have consistently underestimated. The difference between espionage and influence operations is much smaller than people want it to be when you're already inside the system. Vietnam is also, and this is important context that rarely gets enough attention, a case study in a country trying to manage a genuinely complicated relationship with China while maintaining its own sovereignty. Hanoi has real economic interdependence with Beijing. It also has real territorial grievances. It has tried, imperfectly, to thread that needle through a policy of hedging: building relationships with the United States, Japan, India, and others without fully committing to any alignment that would force a choice. Chinese intelligence operations against Vietnamese infrastructure are in part about understanding where that hedging is going. What concessions is Vietnam considering? What are officials saying privately that they're not saying publicly? What's the actual state of Vietnam's relationships with Washington and Tokyo at the working level? That intelligence function doesn't go away. It intensifies as the strategic competition in the region intensifies. The South China Sea has gotten considerably more contested since 2012, not less. The HYSY 981 standoff was a preview of a playbook Beijing has run repeatedly since then, move a large piece of equipment into disputed waters, hold it there, and generate a crisis that gives you leverage. Each time that playbook runs, the cyber operations that accompany it are looking for the same things: what does the target government actually know, what are they planning to do, and are there internal disagreements that can be exploited? For Vietnamese officials and infrastructure defenders, the lesson from 2012 that still applies: the timing of intrusion attempts correlates with diplomatic and territorial events. When Beijing is applying pressure in the physical domain, the intrusion tempo goes up. That's your threat indicator. It's not a coincidence and it's not opportunism. It's synchronization. Which means when you see Chinese government vessels doing something provocative in the South China Sea, your SOC should be paying closer attention, not the same amount of attention. For the broader regional picture, the 2012 campaign against Vietnam was one node in a much larger effort that was hitting, simultaneously or in close sequence, energy and communications targets across Southeast Asia. The Philippines, Indonesia, Malaysia: all facing variants of the same operational template. The actors weren't always the same groups, but the strategic logic was consistent. Map the energy resources, understand the communications, monitor the media, time it to the diplomatic pressure. What that means now is that any serious regional security architecture needs to treat cyber intrusions against energy and communications infrastructure as part of the same problem set as maritime territorial disputes. They are not separate issues being handled by separate bureaucracies in Beijing. They shouldn't be treated as separate issues by the governments defending against them either. One more thing. The Vietnam News Agency and Vietnam Post targeting points to something that rarely gets adequate attention in retrospectives about campaigns like this: the value of access to institutions that aggregate communications. News agencies receive information from a wide range of sources. Postal systems route correspondence between government offices, between officials and their contacts, between businesses and their counterparts. The intelligence value of those aggregation points is not about any single message. It's about the pattern of who is talking to whom, about what, at what frequency, and when the pattern changes. That's the kind of data that makes every other intelligence product more useful. It's map and key at the same time. The 2012 campaign didn't end in 2012. The access, if it was maintained, was the beginning of something. The technical indicators that allowed attribution were a learning opportunity for the operators. The targets that were hit have not stopped being strategically valuable. Twelve years later, PetroVietnam is still drilling, the South China Sea is still disputed, and the communications infrastructure of a country trying to carefully manage its relationship with a much more powerful neighbor is still a collection priority. This is what a sustained, institutionalized intelligence campaign looks like in its early years. Patient. Coordinated with other instruments of statecraft. Focused on the same underlying strategic questions year after year. The specific intrusion methods evolve. The questions being asked don't change much at all. foreigninterference.org/post… #foreigninterference #CyberEspionage #RegionalIntelligenceGathering

Kaspersky's GReAT team disclosed Operation FlashHole last month, a Lazarus Group campaign hitting financial institutions and tech companies that had been running for months before anyone caught it. "Black October" is what SecurityToday called the broader convergence of APT activity that got documented all at once. The name is a little dramatic, but the underlying reality isn't. Here's what the timing tells you. The October disclosures weren't a spike in attacks. They were a spike in visibility. The operations were already running. Lazarus and peer actors had months of uncontested access while defenders were looking at clean dashboards. That gap, between when an intrusion starts and when someone notices, is where the real story lives, and it keeps getting glossed over in favor of the theatrical disclosure moment. Lazarus has been doing this for years. The group's dual mandate is well documented: collect intelligence for Pyongyang, steal money to fund the weapons programs that sanctions are supposed to starve. The cryptocurrency hauls over the past five years run into the billions. FlashHole fits the same template. Financial sector targeting, extended dwell time, patience. They're not in a hurry because they don't have to be. The dwell time question is the one worth sitting with. When an intrusion runs for months before detection, the attacker isn't just exfiltrating files. They're mapping the network, identifying privileged accounts, learning communication patterns, figuring out what's worth taking and what's worth leaving for later. By the time an incident response team kicks the door open, the adversary has had a semester's worth of education on the target environment. The data that's gone is gone, obviously, but the pre-positioning for potential follow-on activity is the part that doesn't show up in the breach notification. Where this is heading: Lazarus keeps refining the patient approach because it works. Living-off-the-land tradecraft, which Kaspersky flagged as dominant in the October landscape, is specifically designed to be invisible to the signature-based tools that most organizations still rely on as their primary defense. Behavioral detection is the answer defenders have been talking about for a decade, but deployment is slow, tuning is hard, and alert fatigue is real. Lazarus knows this. Their operators are reading the same industry reports everyone else is. The next phase isn't going to look dramatically different from FlashHole. That's the point. Supply chain positioning, credential harvesting through legitimate service abuse, long-running footholds in financial infrastructure. The group has demonstrated no pressure to change a method that consistently delivers months of undetected access. Why would they? Sanctions mean Pyongyang needs the revenue. Revenue requires the cyber theft operations to keep running. The strategic incentive structure isn't shifting. What should concern financial institutions specifically: the FlashHole targeting pattern isn't random. North Korea has systematically worked through the stack of institutions that touch cryptocurrency custody, exchange operations, and cross-border transfer infrastructure. A target that thinks it dodged this campaign because it wasn't in the disclosed victim set should probably not feel reassured. The disclosed victims are the ones that got found. The detection gap cuts both ways. For technology companies in the mix, the intelligence collection angle matters separately from the financial theft. Lazarus operates inside a broader North Korean cyber architecture that includes other units focused on different objectives. A tech company compromised for revenue generation purposes can become infrastructure for an operation with different goals entirely. The initial access is the scarce resource. Once it's established, it gets used. The Kaspersky disclosure also surfaces a structural problem that doesn't get enough attention: the gap between when private sector threat intelligence firms have high-confidence attribution and operational detail, and when that information actually gets to the defenders who need it. GReAT is presenting at a security conference. The institutions that were targeted months ago may or may not have gotten actionable warning in time to matter. Intelligence sharing between private researchers and government defenders has improved over the past several years, but "improved" is doing a lot of work in that sentence when the operational timelines still run to months. Financial sector security teams should be specifically auditing for the behavioral indicators that survived the October disclosure window. Kaspersky's conference presentation gives you the shape of what FlashHole looked like. That's the floor, not the ceiling. The next iteration will have absorbed whatever operational lessons the group drew from getting caught, which means the disclosed indicators are already partially obsolete. Government officials watching North Korea's sanctions evasion picture need to account for the fact that the cryptocurrency theft pipeline is funding something. The weapons programs don't pause while the cybersecurity community debates detection timelines. The financial sector compromise isn't a separate story from the geopolitical one. It is the geopolitical one, running through a different department. The "Black October" framing will fade. The operational pattern won't. foreigninterference.org/post… #foreigninterference #AdvancedPersistentThreatOperations #CyberEspionage #PersistentNetworkInfiltration #FinancialCyberTheft #CryptocurrencyTheft #LivingOffTheLandExploitation

Nigeria runs 220 million people and an election in 2027. The EU just flagged it as a priority target for AI-generated disinformation, which is notable given Brussels usually stays in its own lane. INEC has fact-checkers. Generative AI has scale. Those two things are not close to equivalent, especially in Hausa, Yoruba, or Igbo, where detection tools barely function. No named actor in the EU assessment. Doesn't need one. The tools are cheap enough now that the attribution question is almost beside the point. foreigninterference.org/post… #foreigninterference #DisinformationCampaigns #ElectionInterference #DeepfakeVideoCreation #AIProfileGeneration #ComputationalPropaganda

X suspended more than 13,000 accounts linked to foreign influence operations ahead of the 2026 midterms. That number sounds significant until you understand the underlying math: platform takedowns consistently lag behind new account creation, which means the suspension figure tells you more about how fast Russian-linked networks are spinning up accounts than it does about how effectively anyone is stopping them. NPR's reporting on the final stretch of the midterm campaign documents what researchers are calling a refined version of the 2024 playbook. Not a new playbook. A refined one. That distinction matters because it means these operations are getting cheaper and harder to catch at the same time. Two specific changes stand out from what researchers told NPR. First, content is increasingly AI-generated. The old problem with Russian-origin social media content was that it had detectable tells: slightly off phrasing, odd syntax, the kind of thing that native speakers flag immediately. AI generation is eroding that detection advantage fast. The linguistic fingerprint that helped researchers and platforms identify this stuff is becoming much less reliable, which means one of the primary signals analysts depended on is degrading in real time. Second, the operations are now running on two tracks simultaneously. There's still the broad-scale narrative amplification that characterized 2016 and 2024, the kind designed to push divisive content to large audiences and let domestic polarization do the rest of the work. But alongside that, researchers identified a more targeted tier of operations focused on specific candidates, specific congressional districts, and specific demographic groups identified as pivotal to contested races. That's operationally more sophisticated than blasting inflammatory content at everyone and hoping it lands somewhere useful. Narrowing the targeting means resources go toward the places where a marginal shift in sentiment or turnout is most likely to affect an outcome. The combination of those two changes, AI generation reducing production costs and a tiered targeting structure improving efficiency, means Russian-linked operations in 2026 are producing more content, more credibly, aimed more precisely, than in previous cycles. The defensive architecture hasn't kept pace with that. On the federal side, Semafor's reporting describes a specific tension that doesn't have an easy resolution. The Trump administration has publicly positioned itself as prepared for election meddling. At the same time, the federal infrastructure built to counter foreign interference has been significantly reduced, including the drawdown of CISA's election security program. The intelligence community is still generating warnings, but according to Semafor's reporting, those warnings are increasingly going through unofficial channels because the formal architecture that used to carry them has been diminished. Researchers and Democrats characterize this as a deterrence gap. Whether you accept that framing or not, the structural fact underneath it is real: the agencies and programs that existed to coordinate counter-interference responses at the federal level are smaller and less active than they were during the 2020 and 2024 cycles. That matters for a reason beyond the immediate midterm context. Deterrence in the influence operation space is partly a function of whether a foreign actor believes there will be a meaningful response to what they're doing. If the counter-interference architecture is visibly diminished, the implicit signal to operators running these campaigns is that the cost-benefit calculation has shifted in their favor. Meta deployed election integrity measures including reduced political content recommendations in the lead-up to the midterms. That's the standard platform response at this point, and it's not nothing. But reduced recommendations don't address coordinated inauthentic behavior directly, and the structural asymmetry between state-backed account creation and reactive platform enforcement isn't something any single policy adjustment solves. Russia has more patience for this than platforms have budget for manual review. What the NPR documentation, combined with the congressional intelligence disclosures reported by CBS News and Nextgov, gives you is the most detailed open-source picture of Russian 2026 midterm operations available before votes are actually cast. The picture it assembles is of an operation that has taken four years of post-2022 refinement and emerged more capable on the dimensions that are hardest to defend against: production scale, linguistic credibility, and targeting precision. The defensive side of this, federally and at the platform level, is not keeping pace with those improvements. That's not a prediction. That's what the current architecture looks like measured against the documented operational capacity on the other side. foreigninterference.org/post… #foreigninterference #DisinformationCampaigns #ComputationalPropaganda #ElectionInterference #AIProfileGeneration #InfluenceOperations #PlatformTakedown #SocialMediaHijacking

Jim Himes and Mark Warner went public this week with something that would have sounded paranoid four years ago: they've been war-gaming it. Formal contingency planning, working through legislative responses to a range of scenarios before the midterm results are even in. One of those scenarios is that the President of the United States claims foreign interference to contest or delegitimize the outcome. That's where we are. The disclosure came in the final days before the 2026 midterms, reported by CBS News, Nextgov, and the Kyiv Independent. Himes, ranking member on the House Intelligence Committee, and Warner, his Senate counterpart, described an active intelligence community briefing process on Russian and Chinese influence operations targeting the vote. They didn't just describe the threat in general terms. They got specific, which is itself notable, because they're choosing to say this out loud right now. On Russia: the operations they described aren't the blunt "sow discord, burn it all down" approach from 2016 or even the 2024 playbook. What the IC briefed them on this cycle is more operationally precise. Russia is targeting Democratic candidates and sitting members of Congress who support U.S. assistance to Ukraine. Specific people. Specific policy positions. Disinformation operations calibrated to individual vulnerabilities rather than just flooding the zone with chaos. That's a meaningful evolution in tradecraft, or at least in targeting philosophy. Generic discord amplification is cheap and noisy. Candidate-specific targeting takes more effort and suggests clearer objectives: not just destabilizing confidence in elections broadly, but shaping who wins particular races and, by extension, shaping the congressional coalition on Ukraine aid. Warner and Himes were careful to distinguish the Russian and Chinese operations from each other. China's efforts are apparently active but more surgical, focused on specific policy outcomes and candidate vulnerabilities rather than the systemic narrative manipulation Russia specializes in. This tracks with how the intelligence community has long characterized the difference in priorities: Russia wants to delegitimize institutions, make Americans trust nothing, burn the house down. China wants leverage. China wants policy outcomes, trade concessions, less scrutiny of its activities in the Pacific. Different goals produce different methods, and the IC assessment here lines up with years of prior analysis on that distinction. Now, the war-gaming. The formal contingency planning Warner and Himes described is genuinely novel as a disclosed institutional practice. Congressional leaders preparing legislative response frameworks for election interference scenarios is one thing. But they're explicitly including in that planning the scenario where the executive branch itself becomes a variable, specifically the scenario where Trump alleges foreign interference as a post-election narrative tool regardless of what the IC actually finds. That's a very specific thing to plan for. It means Democratic leadership on the intel committees has looked at 2020, looked at the incentive structures currently in place, and decided that treating a presidential misuse of foreign interference allegations as a realistic contingency is just responsible planning. They're not wrong to plan for it. The allegation of foreign interference is remarkably versatile as a political tool. It can be used to cast doubt on an unfavorable result. It can be used to justify delays, investigations, or simply to keep a losing outcome in rhetorical dispute for months. The IC briefings that would normally inform any such claim now exist inside an executive branch that has, in the current administration, shown it's willing to reshape intelligence products to fit preferred narratives. So congressional Democrats having their own frameworks ready, their own response architecture, is a reasonable hedge against a situation where the institutions that are supposed to adjudicate these claims have been compromised or politicized. The optics of announcing this before the results are in are worth thinking about. On one hand, it's transparency, and it puts the planning on the record before anyone can claim it's purely reactive spin. On the other hand, Republicans will characterize it as Democrats pre-emptively refusing to accept results, which is the mirror image of the actual concern but will work fine as a talking point. Himes and Warner presumably did the calculation and decided that the deterrence value of public disclosure, and the institutional record it creates, outweighs the attack surface it opens. Separately and somewhat more comedically, the Ambassador Pete Hoekstra situation has been resolved, at least for now. Earlier this week Hoekstra had raised eyebrows by accusing Canada of interfering in the midterms, citing Prime Minister Mark Carney's public statements on U.S. tariffs. The accusation was always a bit strange: a foreign leader commenting on trade policy is not typically what the interference-tracking community has in mind when we talk about election interference. Politico reported that Hoekstra has since softened that position, now characterizing Carney's statements as political commentary rather than formal interference. A diplomatic de-escalation, then, after a week of bilateral tension that included formal Canadian diplomatic guidance to its officials to avoid appearing alongside U.S. candidates. Canada formally advising its diplomats not to be photographed with American political candidates is, in its own way, a sign of where the relationship sits right now. That's not normal great-power neighborly conduct. That's two countries actively managing the optics of proximity to each other's electoral processes, which tells you something about how badly the relationship has frayed. Carney commenting on tariffs is Carney doing his job. The fact that Hoekstra floated it as interference, even briefly, is either a trial balloon for something or just reflects how broadly the current administration is willing to define the concept when it's useful. The combination of all this, the Russia candidate-targeting, the China surgical influence ops, the congressional war-gaming, and the ambassador's brief foray into calling Canada a foreign interference threat, paints a picture of an election environment where the interference threat is real and the political incentive to weaponize the concept of interference is also real. Those two things coexisting makes the analytical job harder and the institutional response more important. The fact that Warner and Himes built their planning around both problems simultaneously suggests they understand that the threat isn't just foreign. foreigninterference.org/post… #foreigninterference #DisinformationCampaigns #ElectionInterference #InfluenceOperations #CrossBorderInfluenceOperations #DemocraticInstitutionTargeting #IntelligenceAssessmentDisclosure

Russia ran a deepfake smearing Texas state legislative candidate James Talarico as a pedophile in the final days before the midterms. Politifact confirmed it. Bot networks pushed it. This is on the record now. Focus on what it tells us about where this is heading. The targeting was surgical. Not a narrative. Not a meme ecosystem. A specific candidate, a specific charge designed to be unfalsifiable in the gut even after it's debunked in the press. The Kremlin's influence shops have done the math on human psychology: a child abuse allegation lands differently than "this politician is soft on X." You can argue policy. You can't fully un-ring that bell. The 2024 cycle was still mostly vibes-level stuff. Broad wedge content, amplified outrage, accelerant on existing fires. What's different here is the precision. Intelligence briefings to congressional Democrats (NPR and Nextgov both reported on these) confirmed that Kremlin-backed groups were running candidate-specific operations in 2026, not just ambient noise. Talarico fits the profile: Democratic, state-level, pro-Ukraine. Not a national figure with a full rapid-response operation behind him. That's the tell. They're not going after the senators with fifteen comms staffers and standing media relationships. They're going after the candidates who don't have the infrastructure to punch back fast enough. A state leg race. Days before the vote. Deepfake deployed. Even if fact-checkers catch it, the timing was designed so that "Politifact says it's fake" runs after a significant chunk of people have already seen the video. Here's the trajectory. The cost of producing a convincing deepfake has collapsed. What required serious technical lift in 2022 is now accessible enough that even mid-tier influence operations can run it. That means the volume is going to increase. Not by a little. Researchers flagged before the 2026 cycle closed that the total output of AI-generated disinformation in the final campaign stretch would exceed fact-checking capacity. Talarico's case got flagged because enough people raised alarms. The next ten cases may not. What the next cycle looks like: more of this, but spread thinner and deployed faster. Expect it to hit candidates at the city council and state house level who have almost no public profile, meaning less pre-existing context for voters to weigh the claim against. Expect the content to get better. Deepfakes from early 2025 that looked off around the eyes and mouth are already being outpaced by generation tools that have trained on far larger datasets. The gap between "convincing enough to require fact-checking" and "convincing enough to pass casual scrutiny" is closing. The distribution side matters too. Bot networks amplified this one, which is at least partially detectable. The next evolution is seeding this kind of content through what look like authentic accounts, regional Facebook groups, local community forums. Slower burn. Harder to attribute. By the time attribution happens, the election is over. For officials: the fact that this was documented and confirmed is genuinely useful, but the response infrastructure doesn't match the threat timeline. Politifact's intervention was reactive. It happened after the video spread. What's needed is platform-level detection with teeth, not voluntary commitments, and a direct rapid-notification channel from fact-checkers to campaigns being targeted. Neither of those exists at scale right now. For candidates running in the next cycle, especially Democrats in competitive state-level races and anyone with a visible pro-Ukraine record: assume you are a potential target. Have a plan before the video drops, not after. That means pre-built relationships with platform trust-and-safety teams, a rapid-response protocol, and baseline documentation of your actual record on anything that could be weaponized. The attackers are thinking about this. Most campaigns aren't. For voters: the specific charge here, pedophilia, was chosen because it provokes an emotional response that bypasses normal skepticism. That's deliberate. The heuristic for the next two years should probably be: the more viscerally outrageous the last-minute video, the more skepticism it warrants. That's backwards from instinct, which is exactly why it works. Russia's influence operations did not get less sophisticated after 2024. They got cheaper to run, more precise in targeting, and better at timing. The Talarico operation is a proof of concept for the next phase. foreigninterference.org/post… #foreigninterference #DeepfakeVideoCreation #DisinformationCampaigns #ComputationalPropaganda #AIProfileGeneration #ElectionInterference #IndividualTargeting

UEFI bootkits are not new. LoJax was documented in 2018. CosmicStrand and MosaicRegressor came later. What's new is that this capability is apparently now routine enough to be deployed against mid-tier diplomatic targets in Central Asia, which is a different statement about proliferation than most coverage acknowledges. BlackX is the name researchers have given a firmware-level implant currently being used against diplomatic organizations across Central Asia. The technical mechanism is specific and worth understanding: it doesn't exploit some exotic zero-day in the firmware itself. It exploits gaps in Secure Boot's certificate revocation infrastructure. Secure Boot is supposed to ensure only cryptographically verified code loads at startup. But the revocation lists that determine which certificates are no longer trusted have to be actively updated and applied. BlackX targets systems where those revocation updates haven't been pushed. Fully patched at the OS level. Vulnerable at the firmware level. That distinction matters enormously for how defenders think about their update posture. The persistence model is what makes this genuinely hard to respond to. Standard incident response assumes you can wipe and reload. Reinstall the OS, restore from backup, you're clean. Not here. The implant lives in the system firmware, below the OS entirely, so reinstalling Windows does nothing. You need to reflash the firmware. That's a non-trivial operation under any circumstances, and for a diplomatic mission in Bishkek or Dushanbe with limited IT support and probably no in-house security team, it may be effectively impossible without specialized assistance. The attacker's persistence survives most realistic response actions, which is presumably the point. Central Asia as a targeting geography makes sense when you look at what's actually there. The region hosts diplomatic missions from NATO members, EU states, and a range of multilateral organizations. Given the current environment, those missions are actively working through questions about the war in Ukraine, sanctions policy, and the various connectivity and infrastructure initiatives competing for regional influence. Russian and Chinese intelligence services both have obvious collection interests there. Western services do too. Attribution on BlackX hasn't been publicly confirmed, and that's worth taking at face value rather than speculating past what the evidence supports. The proliferation angle deserves more attention than it usually gets in coverage of individual campaigns. LoJax was attributed to Fancy Bear and caused genuine alarm when it surfaced in 2018 precisely because UEFI implant capability seemed like a high bar. CosmicStrand and MosaicRegressor extended that picture to Chinese-linked actors. BlackX continues the pattern, but the more significant observation is structural: firmware-level persistence is no longer a capability that signals only the most elite tier of state actor. It's becoming a standard tool in advanced persistent threat operations. The attack surface exists on virtually every modern computer. The exploitation technique here, abusing revocation gaps rather than requiring novel firmware vulnerabilities, is reproducible. Defenders who haven't thought carefully about firmware update management and revocation policy are operating with an assumption about their security baseline that may not hold. foreigninterference.org/post… #foreigninterference #AdvancedPersistentThreatOperations #KernelLevelRootkitImplantation #ZeroDayExploitation #CyberEspionage #PersistentNetworkInfiltration #UefiFirmwareImplantation

Zhu Rongji went on state television weeks before Taiwan's 2000 election and told voters not to make the wrong choice. No cutouts, no plausible deniability, just a premier of a nuclear-armed state lecturing a democracy about who it was allowed to elect. Chen Shui-bian won anyway. Beijing has spent the decades since refining the approach, but the underlying logic never changed. foreigninterference.org/post… #foreigninterference #ElectionInterference #DiplomaticCoercion #StateMediaCoordination #MilitaryIntimidation

The DNI's CI milestones archive marks March 15, 2001 as the date David Szady became the first National Counterintelligence Executive. Useful history. The more useful part is what it tells you about where we're going now. Szady's appointment didn't come from nowhere. It came from Ames. It came from Hanssen. Two men who between them handed Moscow decades of penetration, largely because the FBI and CIA were too busy not talking to each other to notice what was walking out the door. The NCIX position was the institutional response: centralize, coordinate, produce annual threat assessments, build a bridge between agencies that had spent the Cold War treating each other as bureaucratic rivals almost as threatening as the Soviets. Here's the part that should give you pause. The reform came late. Years late. The Ames arrest was 1994. Hanssen wasn't rolled up until February 2001, three weeks before Szady's appointment. That's seven years of documented catastrophe before the coordination architecture got its first dedicated executive. Seven years during which both agencies knew something was wrong and couldn't figure out how to fix it together. That lag is not a historical curiosity. It is a pattern that has repeated with enough consistency that you can set your watch by it. After 9/11, the lag between failure and institutional response was roughly two years. The 9/11 Commission report published in 2004, the IRTPA passed that same year, the DNI stood up in 2005. Information-sharing failures between FBI and CIA that people had flagged internally before the attacks finally got addressed after the attacks. The 2016 election interference cases followed a similar arc: the Senate Intelligence Committee's final report ran to five volumes and took until 2020. The NCSC, which is the successor to the NCIX that Szady first led, only got serious public-facing election security infrastructure going in 2018, two years after the most significant foreign interference operation in modern American political history. The lag isn't stupidity. It's institutions. Bureaucracies don't reorganize around threats they haven't yet experienced at scale. They reorganize around the last disaster. That's not cynicism; that's how large organizations actually work. The problem is that foreign intelligence services have figured this out and they operate on the assumption that the window between their next significant operation and the American institutional response will be at least eighteen months to three years. Possibly longer if the operation is politically ambiguous enough that there's a domestic argument about whether it even happened. So what does the Szady appointment, and the NCIX architecture it spawned, actually tell us about trajectory? First, watch what the NCSC is being asked to do right now, and watch what it isn't being resourced to do. The NCSC is the direct descendant of what Szady started. It produces threat assessments, coordinates across agencies, publishes advisories. It is also, by design, a coordination body, not an operational one. It doesn't run cases. It doesn't arrest people. It depends entirely on the FBI and CIA actually sharing information with it and acting on its assessments. When the relationship between those two agencies gets strained, or when political pressure discourages the FBI from treating certain foreign influence operations as priority threats, the NCSC's institutional position doesn't give it much leverage. Szady's original mandate had the same limitation. The Counterintelligence Enhancement Act of 2002 gave the NCIX statutory grounding but not independent authority. The successor structure inherited that constraint. Second, the NCIX's 2009-2011 reporting on Chinese and Russian economic cyber espionage is worth revisiting not for what it said but for what happened after it was published. Those reports were accurate. They documented systemic, large-scale theft of American intellectual property and trade secrets. They named the actors. They went to Congress. And for roughly three years after publication, the primary response was diplomatic complaint and private sector awareness campaigns. The operational and legal infrastructure that eventually produced the DOJ's China Initiative didn't really materialize until 2018, and the China Initiative itself was shut down in 2022 amid concerns about civil liberties and effectiveness. Whatever replaces it is still being built. The gap between knowing and doing is where foreign intelligence services live. They count on it. Third, and this is the forward-looking part: the NCSC right now is trying to do something Szady's original mandate never contemplated at scale, which is addressing foreign interference in the information environment. The 2001 NCIX was built around the espionage threat model: recruited agents, stolen secrets, compromised programs. Classic tradecraft. The election interference problem and the broader foreign influence operation problem don't fit that model cleanly. There's no Robert Hanssen to arrest. The operations run through social media platforms, through influence networks, through legitimate-seeming political activity, sometimes through people who aren't recruited agents but are useful idiots or witting amplifiers. The legal authorities, the investigative frameworks, the interagency coordination mechanisms all evolved for the old model. What this means in practice: the NCSC can identify and attribute foreign influence operations with reasonable confidence. The 2024 election cycle produced multiple attribution statements about Iranian, Russian, and Chinese activity. The gap is between attribution and consequence. What happens after the NCSC publishes an advisory? The platforms may or may not act. Law enforcement may or may not have a legal hook. Congress may or may not hold a hearing. The foreign actor almost certainly keeps operating because the cost-benefit calculation still works in their favor. That gap is going to widen before it narrows. Here's why. The tools that make influence operations cheap and scalable are getting cheaper and more scalable. Generative AI is not a theoretical threat to the information environment; it's already being used. The Iranian operation targeting the 2024 election used AI-generated content. That's documented. The tradecraft will evolve faster than the legal and institutional responses can, because the legal and institutional responses require consensus, legislation, interagency coordination, and time. The actor just needs a server and an API key. The NCSC's current advisory model assumes a relatively stable information ecosystem where you can identify foreign content, label it, and let platform moderation handle the rest. That model is already strained. It breaks entirely if the volume of synthetic foreign-origin content reaches a point where moderation at scale becomes impossible, or if the content becomes indistinguishable from domestic political speech even to expert reviewers. Neither of those scenarios is far off. What should defenders watch for? A few specific things. Platform-specific targeting is going to get more granular. The 2016 operations were relatively blunt: broad reach, high volume, demographic targeting that any competent data analyst could reverse-engineer after the fact. More recent operations show evidence of narrower targeting, focusing on specific communities, specific local issues, specific electoral districts. The Iranian operation in 2024 showed more targeted outreach, including direct contact with campaign officials. The trajectory is toward precision over reach. Smaller operations, harder to detect, focused on outcomes rather than exposure metrics. The use of legitimate-seeming American intermediaries is increasing. This is the thing that doesn't fit the old NCIX espionage model at all. Foreign actors have learned that content laundered through American voices, American organizations, or Americans who simply don't know they're amplifying foreign-origin narratives is vastly more effective and legally harder to address than content that's obviously foreign. The DOJ has brought cases on this, mostly under FARA. The cases are hard. The legal standard is high. The actors have adjusted accordingly. State and local targeting is the next frontier and it's largely undefended. Federal counterintelligence infrastructure is oriented toward federal targets. The NCSC coordinates with state election officials, but the information security and counterintelligence capacity at the state and local level is, with some notable exceptions, minimal. Foreign actors have noticed. Targeting local officials, local media, local political organizations is cheaper, easier, and currently subject to less scrutiny than targeting federal institutions. The Szady-era NCIX had essentially no mandate in that space. The NCSC's mandate is broader but its resources don't match the scope of the problem. For officials: the 2001 reforms didn't prevent 2016. The post-2016 reforms are not going to prevent the next significant operation, whatever form it takes. The goal isn't prevention; it's detection speed, response capacity, and cost imposition. Right now the cost imposition piece is the weakest link. Sanctions exist. Indictments exist. Neither changes the calculus enough to deter well-resourced state actors. Until that changes, the rest is largely damage management. For voters: attribution statements from the NCSC are not partisan spin. They're the product of an institution that has existed in some form since March 2001, that survived administrations of both parties, and that has a statutory mandate to be accurate because it testifies to Congress. When the NCSC says a foreign government is trying to influence an election, that's worth taking seriously. It's also worth understanding that their public statements are the sanitized version of what they actually know. Szady took over a coordination function that was built on the wreckage of Ames and Hanssen. The institution that grew out of his appointment is now trying to address threats that nobody in March 2001 had the vocabulary to describe accurately. That's not a criticism. It's a description of how counterintelligence always works: reactive, structural, slow. The adversaries are patient and they plan for the lag. The question is whether the institutions can close that gap before the next significant failure makes it politically mandatory to try again. foreigninterference.org/post… #foreigninterference #IntelligenceRestructuring #CounterintelligenceOperations #OversightReform

Late October 2026, and Sandworm was back in Ukrainian infrastructure. That part alone would be news. What made this operation different was who else showed up. Researchers documented a joint attack on Ukrainian critical infrastructure that paired Sandworm, the Russian military intelligence hacking group responsible for some of the most destructive cyberattacks in history, with a second, still-unidentified threat actor. Together they deployed a wiper payload alongside something that hadn't been seen before: a mesh network backdoor. The combination is worth understanding on its own terms, because it changes the defensive calculus in ways that matter beyond Ukraine. Start with the mesh architecture, because it's the part that most shifts the threat. Conventional command-and-control infrastructure for this kind of operation works on a hub-and-spoke model. There's a central server, compromised machines check in, defenders find the server or block the traffic and the operation is disrupted. It's not easy to do, but the model is understood. A mesh topology throws that out. Compromised nodes relay instructions to each other laterally. Take one node offline, and the network routes around it. Defenders trying to disrupt the operation can't just sever a central connection, because there isn't one. They have to find and clean every node, which in a large enterprise or critical infrastructure environment means chasing a problem that keeps regenerating itself across the network. Pair that with a data wiper and you have an operation designed to degrade and destroy while remaining present. The wiper does the immediate damage. The mesh backdoor ensures operators can come back, push new payloads, or maintain visibility even after defenders think they've contained the incident. In attacks against critical infrastructure, the timing of that return access matters enormously. An energy facility or data center that believes it has expelled the attacker and begins recovery operations, only to have the attacker still present and watching, is in a far worse position than one that knows the fight is ongoing. The second threat actor is a different kind of problem. Russian intelligence has been running this playbook for years, pulling in criminal groups and quasi-independent hackers to expand capacity and create distance from the GRU or FSB fingerprints that Western agencies can attribute with reasonable confidence. Sandworm itself has been linked directly to the GRU's Unit 74455. When a known state actor operates alongside an unnamed partner, that partner could be criminal, could be another intelligence service element, could be a contractor arrangement. The plausible deniability that arrangement creates is real, even if Western analysts see through it most of the time. It also expands what Sandworm can do simultaneously. Two actors with different toolkits, operating in coordination, can hit more targets or hit them from more angles than a single group working alone. The NotPetya comparison is relevant here and worth stating plainly. In June 2017, Sandworm deployed what looked like ransomware but was actually a wiper masquerading as ransomware, using a tax software update in Ukraine as the initial vector. It spread globally, hit shipping companies, pharmaceutical manufacturers, hospitals, and port operators across Europe and North America. Total estimated damages reached $10 billion. Nobody asked for that spillover. It happened because destructive tools built for use in Ukraine don't stay in Ukraine. That's the consistent pattern. Capabilities tested in Ukrainian networks migrate. CISA and allied agencies have said it explicitly in multiple advisories over the past several years. The mesh backdoor technique documented in this operation is new enough that defenders haven't had time to develop robust detection signatures for it, and the research publication is expected to prompt CISA and partner agencies to release technical indicators. That matters, but there's a lag between a novel technique appearing in the wild and defenders having the tools and training to catch it. That lag is where the danger sits. The timing of the October operation fits Russia's established seasonal rhythm. In the years since the full-scale invasion began in February 2022, Russian destructive cyber operations against Ukrainian infrastructure have intensified in the months before winter, coordinated with or running parallel to missile and drone strikes on power generation, heating systems, and communications. Data centers have become explicit military targets in this conflict. That's not a rhetorical point. Both sides have treated information infrastructure as legitimate wartime objectives, and Russia in particular has tried to degrade Ukrainian data processing and communications capacity as part of a broader campaign to make the country harder to govern and harder to defend. The mesh backdoor makes that kind of operation harder to eject and harder to monitor. Defenders who think they've cleared a compromised network and restore operations may be restoring operations with an attacker still inside. The wiper component creates immediate, visible damage. The mesh component creates persistent, less visible access. Used together against the same targets, they're designed to maximize both immediate disruption and long-term presence. What the research doesn't yet resolve is the identity of the second threat actor and the precise division of labor in the operation. Those questions matter for attribution and for understanding how Russian intelligence is structuring its outsourcing arrangements. The operational collaboration model itself, a known state actor working alongside a less-attributable partner, is not new. But each new instance of it provides insight into how Russia is managing the tension between wanting deniable capacity and wanting coordinated, sophisticated operations. The two goals pull in opposite directions. Deniable partners are harder to trust and harder to control. The fact that this operation appears to have been executed with enough coordination to combine a novel backdoor with a wiper payload suggests the collaboration was structured and probably not improvised. Western cyber defenders should be watching the technical indicators when CISA releases them and mapping their own network architectures for the conditions that make mesh-based C2 particularly effective. Flat networks with excessive lateral movement capability, insufficient segmentation between critical systems, and inadequate monitoring of peer-to-peer traffic between internal hosts are exactly what a mesh topology exploits. None of that is new advice. The urgency of acting on it is higher now. foreigninterference.org/post… #foreigninterference #AdvancedPersistentThreatOperations #WiperAttacks #MalwareDistribution #AptCollaborationModel #DataCenterKineticTargeting #CoordinatedCyberWarfare

Academics were writing this up in 2012 and nobody particularly wanted to hear it. The framework was already clear: bots for amplification, social fractures as the attack surface, open information environments as the vulnerability. Authoritarian states could punch way above their weight without a single soldier crossing a border. The platforms had no answer for it, and honestly weren't looking for one. Twelve years of confirmation later, here we are. foreigninterference.org/post… #foreigninterference #CognitiveExploitation #DisinformationCampaigns #SocialMediaHijacking

The U.S. spent decades assuming ideologically motivated spies when the actual driver was almost always money. PERS-TR-92-005 looked at 23 Air Force cases from the late Cold War and found the average detection lag was five to ten years. Most were caught by defectors or accidents, not counterintelligence. The polygraphs weren't working, the financial monitoring barely existed, and workplace culture actively discouraged reporting colleagues. The 1992 study informed a round of reforms. Then Ames ran until 1994. Then Hanssen ran until 2001. foreigninterference.org/post… #foreigninterference #MilitaryEspionage #AssetRecruitment #DeepCoverOperations #PersonnelSecurityEnhancement #CounterintelligenceOperations #IntelligenceSelling

Samuel Loring Morison was not a spy in any conventional sense. He was a Naval Intelligence Support Center analyst who, in 1984, handed Jane's Defence Weekly classified KH-11 satellite photographs of a Soviet aircraft carrier being built. He got paid for it. Jane's published them. And then the United States government decided to prosecute him under the Espionage Act, 18 U.S.C. §§ 793 and 794, statutes that everyone had assumed, for decades, were aimed at people passing secrets to Moscow. That assumption turned out to be wrong, or at least legally unenforceable. The Fourth Circuit upheld the conviction in 1988, ruling in United States v. Morison, 844 F.2d 1057, that nothing in the Espionage Act's text actually limits its application to foreign recipients. If you pass national defense information to someone not authorized to receive it, you're potentially in the statute's reach. Doesn't matter if that someone is a KGB handler or a defense correspondent in London. The court dismissed the First Amendment arguments with similar efficiency: classified information doesn't acquire constitutional protection just because a journalist is the one receiving it. This was the first time in American history that the Espionage Act had been applied to a deliberate media leak. Worth sitting with that for a moment. The statute had been on the books since 1917. Seventy-one years of government officials leaking things to reporters, and no one had gone to prison for it under this particular theory until Morison. The specifics matter here. Jane's Defence Weekly is not some fringe outlet. It was, and is, one of the most respected defense and intelligence publications in the world. Morison wasn't dropping documents in a parking garage with a source he'd cultivated for ideological reasons. He was selling photographs to a publication that covered exactly this kind of military hardware as its core editorial function. There's something almost mundane about the transaction, which is part of what made the prosecution feel aggressive to its critics. And it was aggressive, deliberately so. The Reagan administration was operating in the aftermath of what the intelligence community called the "Year of the Spy." 1985 had produced the arrests of John Walker, Jonathan Pollard, Ronald Pelton, and Larry Wu-tai Chin, among others. The CI disaster that year was real and extensive. Against that backdrop, the administration wanted to signal that unauthorized disclosures would be treated as a category of threat rather than a personnel matter. Administrative sanctions, revoked clearances, that kind of thing. They wanted prosecutions. Morison gave them one. The Fourth Circuit gave them the legal architecture to build on. What the ruling left unresolved is actually where the long-term complications live. The court didn't address whether journalists who receive and publish leaked classified information could themselves be prosecuted. That question sat open, and it has become more pointed over time, not less. The government's theory in Morison logically extends toward the press end of the transaction if you follow it far enough, and subsequent administrations have tested that boundary repeatedly without quite crossing it, at least not yet. The Obama administration used the Morison precedent heavily. The number of Espionage Act leak prosecutions under Obama exceeded the total from all previous administrations combined. Thomas Drake, Stephen Kim, John Kiriakou, Jeffrey Sterling, James Risen's source situation, Edward Snowden charged in absentia. Morison was the foundational case every time. The legal theory that a leak to a reporter is functionally equivalent to a leak to a foreign service, for purposes of criminal exposure, traces directly back to the Fourth Circuit's 1988 analysis. Clinton pardoned Morison in January 2001, which is one of those end-of-term pardons that gets lost in the noise but carries some significance. The pardon doesn't undo the precedent. The conviction stood, was appealed, was upheld, went to the Supreme Court, cert was denied, and it has been good law ever since. Clinton's decision to pardon him reads more as a personal judgment about proportionality than a legal statement about the case's correctness. The constitutional tension the ruling created has never been formally resolved. There is still no statutory distinction in the Espionage Act between a cleared official who sells photographs to a defense publication for money, someone who leaks to the press out of genuine public interest concern, and a foreign intelligence asset. They're all potentially in the same statutory bucket. Congress has never fixed this, despite periodic efforts and a lot of commentary from legal scholars, civil liberties organizations, and press freedom advocates. The classification system and the Espionage Act interact in ways that would require significant political will to rationalize, and that will has not materialized. What the Morison case actually did, functionally, is shift the equilibrium of power over national security information. Before 1988, the practical enforcement mechanism for leaks was administrative. After 1988, it's criminal. That's a significant difference in deterrent effect, and it operates asymmetrically. Senior officials who leak selectively to shape coverage generally face no consequences. Mid-level analysts and contractors who leak without authorization face potential espionage charges. The statute is formally neutral but its enforcement has never been. Morison himself was, in some respects, a poor vehicle for the civil liberties arguments being made on his behalf. He was paid. He held a security clearance and signed the associated agreements. He wasn't a whistleblower raising concerns about government misconduct. He was selling access to classified imagery. But cases rarely arrive at appellate courts wearing their best clothes, and the Fourth Circuit's ruling applied regardless of Morison's particular motivations. The legal principle it established extends to everyone downstream. The KH-11 photographs he leaked were significant. The KH-11 was the United States' primary electro-optical reconnaissance satellite at the time, and imagery from it showing a Soviet carrier under construction provided real intelligence value to anyone analyzing Soviet naval development. The government's classification interest was legitimate. But the court's ruling establishing that this legitimate interest could be vindicated through criminal prosecution of a media leaker, using a statute written in 1917 to address wartime sabotage and foreign espionage, is what made the case consequential beyond its facts. Thirty-six years later, the Fourth Circuit's analysis in Morison remains the controlling framework. Anyone thinking seriously about the legal exposure of national security journalists, the prosecution theory in any leak case, or the structural relationship between the classification system and press freedom has to start here. foreigninterference.org/post… #foreigninterference #ClassifiedDocumentLeaking #IntelligenceExposure #WhistleblowerProsecution #LegalWeaponization

Pollard's handlers received the last delivery in 1985. The CIA's damage assessment wasn't finished until 1988, after the sentencing, which tells you something about how long it took to actually map what had left the building. The declassified CIA assessment, released through ISCAP, is worth sitting with because the headline story, Israel running a U.S. Navy analyst, was never really the full story. The operative finding in this document is the third-country compromise. Pollard handed over classified signals intelligence, satellite imagery, and analytical assessments between 1984 and 1985. What happened to that material afterward, inside Israeli intelligence channels, is where the assessment gets structurally interesting and where the redactions get heaviest. The third-country dimension points toward the Soviet Union or Soviet-aligned states. The mechanism isn't spelled out cleanly in the declassified version, but two routes are plausible from what's visible in the document's structure: Israeli intelligence channels that were themselves penetrated, or independent exploitation of vulnerabilities in how Pollard operated. The tradecraft angle matters here. An analyst passing physical volumes of classified material creates a different exposure profile than a digital exfiltration. The material existed in forms that could be re-photographed, copied, shared. Once it left U.S. custody and entered a second intelligence service's holdings, the U.S. had exactly zero visibility into where it went next. The assessment catalogs the damage in ways that compound. Exposure of collection methods and sources across multiple intelligence disciplines is the obvious one. But the CIA also flagged the compromise of allied-nation intelligence that had been shared with the U.S. under classification agreements. That's a second-order problem that doesn't get enough attention: when a partner service shares intelligence with you and your analyst hands it to a third party, you've broken something that isn't just bilateral. The allied nation whose intelligence got passed along had no idea their material was in play. And then there's the analytical priorities dimension. The assessments Pollard passed weren't just raw intelligence. They were finished products that revealed what the U.S. was focused on, which any competent adversary analyst could reverse-engineer to identify where U.S. collection had gaps. That's a gift that keeps giving long after the specific intelligence ages out. The 1988 production date matters for context. The intelligence community was doing this damage assessment while simultaneously trying to reform insider threat detection in the aftermath. Pollard had legitimate access. He wasn't a penetration agent inserted into a target organization. He was already inside, already cleared, already trusted with the specific categories of material he was stealing. The CIA assessment essentially became the canonical template for what that threat profile looks like at scale, which is why it remained relevant to declassify and why the ISCAP release drew attention. The U.S.-Israel intelligence relationship took real institutional damage. The CIA in particular maintained significant reluctance about sharing with Israeli services for years afterward, not because of any formal policy change but because of what you might call an earned wariness at the working level. When an allied service runs an operation against you and the material ends up (probably) in Soviet hands, the damage isn't just operational. It's the entire trust infrastructure that makes allied intelligence sharing work. The formal relationship continued. The actual depth of what got shared contracted. One structural lesson the assessment surfaces, even through the redactions: bilateral espionage operations don't stay bilateral when the recipient nation has its own vulnerability profile. Israel in 1984-85 was operating in an environment where Soviet intelligence services had significant penetration efforts directed at Israeli institutions. The U.S. handing Israel a cleared analyst's output is one transaction. What happens to that output inside Israeli handling chains is a separate exposure surface the U.S. couldn't control and apparently didn't fully account for before the fact. The Pollard case is sometimes treated as a closed chapter, a Cold War artifact. The ISCAP release suggests the intelligence community still sees value in having the damage assessment in the public record, which probably reflects its continued use as a reference document for how insider threats and third-country compromise interact. The lesson isn't complicated, but the Pollard file documents it with more specificity than most cases ever will: access plus ideology plus volume plus an allied service with its own adversaries equals a loss that took three years just to map. foreigninterference.org/post… #foreigninterference #MilitaryEspionage #IntelligenceSelling #AssetRecruitment #DeepCoverOperations #IntelligenceAssessmentDisclosure