UK Govt. Warns on Growing AI Cyber Threats

PDF LINK


#19,122

In July of 2015, in The Lloyd’s Business Blackout Scenario, we looked at the impact of a prolonged grid down disaster brought on by a deliberate cyber attack, which they describe as:

The report depicts a scenario where hackers shut down parts of the US power grid, plunging 15 US states and Washington DC into darkness and leaves 93 million people without power.

Experts predict it would result in a rise in mortality rates as health and safety systems fail; a decline in trade as ports shut down; disruption to water supplies as electric pumps fail and chaos to transport networks as infrastructure collapses.  

A few months later, newsman Ted Koppel published his book Lights Out: A Cyberattack, A Nation Unprepared, Surviving the Aftermath, which warned that our electrical grid was vulnerable to a number of threats, including cyber attack.

Two years later, in DHS: NIAC Cyber Threat Report - August 2017, we looked at the President's National Infrastructure Advisory Council's 45-page report on the urgent cyber threats to our critical infrastructure.

Again in 2018, we looked at another Presidential Advisory report (see NIAC: Surviving A Catastrophic Power Outage), which warned:

https://www.hsdl.org/?view&did=819354

What is a catastrophic power outage?

• Events beyond modern experience that exhaust or exceed mutual aid capabilities
• Likely to be no-notice or limited-notice events that could be complicated by a cyber-physical attack
• Long duration, lasting several weeks to months due to physical infrastructure damage
• Affects a broad geographic area, covering multiple states or regions and affecting tens of millions of people
• Causes severe cascading impacts that force critical sectors—drinking water and wastewater systems, communications, transportation, healthcare, and financial services—to operate in a degraded state

 (Excerpt From Dec 2018 NIAC Report)

In 2022 we looked at two reports issued by the White House on the very real risks of increased cyber attacks against both the private and public sector, which could have profound economic, and societal impacts.

FACT SHEET: Act Now to Protect Against Potential Cyberattacks
MARCH 21, 2022

Statement by President Biden on our Nation’s Cybersecurity
MARCH 21, 2022 

From Ransomware, to distributed denial-of-service (DDoS) attacks - to malicious code which could potentially shut down power plants, interrupt communications, or destroy financial records - tens of thousands of cyber attacks are launched against the United States each day.

While most of these attacks are thwarted, last year the FBI reported that in 2024, their Internet Crime Complaint Center (IC3) reported 859,532 complaints of cybercrime in the US in 2024, resulting in over $16 billion in losses - a 33% increase in losses from 2023.

Fast forward to 2026, and AI (artificial intelligence) apps are already being used to improve the ability of bad actors to hack sophisticated systems (see NBC News Report).  

While primarily used as a `force multiplier' until now, the next generation of AI (aka `Mythos') - announced April 7th by Anthropic - has been deemed by its creators to be too dangerous to release to the general public. 

This has raised considerable concern around the world, as numerous other AI systems are under development, and it is likely that more will achieve this level of sophistication (read:danger) in the months ahead. 

Which brings us to the following UK Government statement on the risks of `Mythos', and the evolving threat from AI tools. 

AI cyber threats: open letter to business leaders (HTML)

Published 15 April 2026

The Rt Hon Liz Kendall MP
Secretary of State for Science, Innovation and Technology
22-26 Whitehall
London SW1A 2EG

The Rt Hon Dan Jarvis MP
Minister of State for the Cabinet Office
70 Whitehall
London
SW1A 2AS

15 April 2026

Dear business leaders,

Open letter to businesses on AI cyber threats

We are writing to you because the threat your business faces in cyber space is changing, and the way we respond must change with it.

For years, the most serious cyber attacks have relied on a small number of highly skilled criminals. That is now shifting. A new generation of AI models are becoming capable of doing work that previously required rare expertise: finding weaknesses in software, writing the code to exploit them, and doing so at a speed and scale that would have been impossible even a year ago.

Last week, AI firm Anthropic announced a new model called Mythos. Testing by DSIT’s AI Security Institute (AISI) - one of the world’s leading bodies for evaluating the capabilities of Frontier AI - has found it to be substantially more capable at cyber offence than any model we have previously assessed. Recent tests of advanced AI models, including the AISI’s evaluation of Anthropic’s Mythos, indicate that AI cyber capabilities are accelerating even faster than had been previously envisaged. The AISI assess that frontier model capabilities are doubling every 4 months, compared to every 8 months previously.

This finding is significant both for what it means today, but also because it highlights the speed at which AI capabilities are increasing and the threats they potentially pose. OpenAI also announced scaling up their Trusted Access for Cyber program last night, showing that AI’s accelerating impact on cyber is not isolated to a single company, and we expect more to follow. The trajectory is clear and therefore it is vital that we are prepared for frontier AI model capabilities to rapidly increase over the next year, and plan accordingly for that outcome.

The UK is not standing still in response to this threat. We have built the AI Security Institute, the most advanced capability of any government in the world for understanding frontier AI systems. This ensures that your government can have an independently verified, robust assessment of current capabilities.

More broadly, the National Cyber Security Centre, part of GCHQ, is world-leading in defending the UK online, and continues to publish practical guidance every business can use. The Cyber Security and Resilience Bill, which is currently progressing through Parliament, will strengthen protections for critical services – from the NHS to the energy system – that we all rely on, and shortly we will publish the National Cyber Action Plan setting out the steps this government will take to ensure the UK’s national security against cyber threats.

Government action alone will not be enough. Every business in the UK has a part of play. Criminals will not just target government systems and critical infrastructure. They will target ordinary companies, of every size, in every sector. Attackers go where defences are weakest.

The steps organisations should take to protect against AI-driven cyber threats are the same cyber hygiene measures recommended for traditional cyber threats. We are asking every business leader reading this to take the following steps:

1. Take cyber security seriously, at the very top of your organisation.

If your board has not recently discussed cyber risk, do so at your next meeting and then regularly. This is not an issue to delegate to your IT team and forget about. This will only become increasingly important. We urge you and your board to use the Cyber Governance Code of Practice to ensure your organisation is sufficiently protected. Smaller businesses should also use the NCSC’s Cyber Action Toolkit to help them build their cyber protection. Not all incidents can be prevented, so you should plan and rehearse how your organisation would respond to a significant incident, including consideration of how cyber insurance can support response and recovery. Free cyber insurance is available to small organisations that obtain Cyber Essentials.

2. Get the basics right with Cyber Essentials.

Most successful cyber-attacks exploit simple weaknesses: outdated software, weak passwords, missing backups. Cyber Essentials is the government-backed certification scheme that protects against the most common attacks. Organisations that hold it are significantly less likely to suffer a damaging cyber incident. For most businesses, getting certified is neither expensive nor difficult. You should also look to embed Cyber Essential requirements across your supply chains, and large organisations should use the NCSC’s Cyber Assessment Framework.

3. Follow NCSC advice and sign up to their Early Warning Service.

The National Cyber Security Centre (NCSC) provides free, practical advice, training and guidance at ncsc.gov.uk, for organisations of every size. Advice will also be issued by Regulators for regulated sectors. Early Warning is a free service from NCSC, which can inform organisations of potential cyber attacks and give them invaluable time to act before an incident escalates.

We are entering a period in which the pace of technological change may test every institution in the country. The businesses that act now – that treat cyber security as an essential part of running a modern company, not an optional extra – will be the ones best placed to thrive through it and seize its advantages. We urge you to be among them.

Yours sincerely,

The Rt Hon Liz Kendall MP
Secretary of State for Science, Innovation and Technology

Dan Jarvis MBE MP
Security Minister, Cabinet Office and Home Office

While the internet seems obsessed with the idea that killer robots, or some all-powerful AI overlord will destroy humanity, the reality is that it will probably be humans - using AI. technology for ill-gotten gain - that pose the biggest danger. 

While there is very little you and I can do to prevent cyber attacks on large industries or our critical infrastructure, we can be better prepared to deal with its potential impacts. 

A few recent blogs on how to prepare for long-term (days or even weeks) power outages, and other disruptions, include:

#NPM: DOE Resource Adequacy Report & Prepping For Power Outages

#NPM25: Preparedness Starts At Home

Denk Vooruit: The Netherlands National Citizen Preparedness Drive

While I can't tell you what disruptions will come, or when they might occur, I can tell you that being prepared - in advance - is the best insurance you and your family can have in an increasingly uncertain world. 

Eurosurveillance: Imported case of Avian Influenza A(H9N2) Virus Infection in a Patient with Miliary Tuberculosis, Italy, March 2026

#19,121

Just over 3 weeks ago (March 25th) we learned of the first confirmed H9N2 infection in Europe, in a traveler recently arrived from Senegal. Initial details were scant, with additional details revealed in the ensuing weeks by the ECDC and WHO DON report. 

On Thursday the Journal Eurosurveillance published the most detailed report to date, which outlined not only their clinical findings and virus characterization; it describes the timeline, and the challenges in identifying the virus. 

As we've discussed often, it can require a bit of luck to accurately diagnose a novel flu infection, and it is assumed that some - perhaps many - go unidentified. 

Novel viruses can often present with only mild-to-moderate symptoms in otherwise healthy individuals, and testing by GPs and clinics are unlikely to differentiate between seasonal and novel flu strains. 

In 2024 the ECDC issued guidance for member nations on Enhanced Influenza Surveillance to Detect Avian Influenza Virus Infections in the EU/EEA During the Inter-Seasonal Period., which cautioned:

Sentinel surveillance systems are important for the monitoring of respiratory viruses in the EU/EEA, but these systems are not designed and are not sufficiently sensitive to identify a newly emerging virus such as avian influenza in the general population early enough for the purpose of implementing control measures in a timely way.

Generally speaking, hospitalized patients with severe symptoms are most likely to receive the type of testing needed to diagnose novel flu. 

In this case, the patient - who presented to the ER severely ill with suspected miliary tuberculosis - initially tested negative for influenza A/B, RSV & COVID from a standard nasal-pharyngeal swab (NPS).

On day two of their hospitalization the patient was subjected to a more invasive BAL (Bronchoalveolar lavage), which confirmed the presence of  Mycobacterium tuberculosis and revealed an untypable influenza A virus.

Although a novel influenza virus was now suspected - and the patient was started on oseltamivir - it would take another 4 days (Day 6) for H9N2 to be confirmed by their National Influenza Centre (NIC).

The full report is very much worth reading, but much of it is technical, and will be of greatest interest to clinicians. I've only posted some excerpts below.  I'll have a bit more after the break.

Imported case of avian influenza A(H9N2) virus infection in a patient with miliary tuberculosis, Italy, March 2026  

Elena Pariani1 , Simona Puzelli2 , Gabriele Del Castillo3,4 , Greta Romano4,5 , Luca Mezzadri6,7 , Cristina Galli1 , Irene Maria Sciabica8 , Luigi Vezzosi3 , Francesca Sabbatini6 , Cristina Paduraru1 , Irene Mileto4,5 , Marcello Tirani9 , Anna Teresa Palamara2 , Paola Stefanelli2 , Fausto Baldanti4,5,10 , Danilo Cereda3,4 , Paolo Bonfanti6,7 , Collaborating Centres’ Study Group on Influenza11

 In March 2026, avian influenza A(H9N2) virus was identified in Italy in a patient with weakened immune system. They had recently travelled to West Africa, which raised concerns about the potential importation of zoonotic influenza viruses into Europe, as H9N2 has been endemic in poultry across the region since 2017, with widespread outbreaks and two human cases reported in Senegal (one in 2020) and Ghana (one in 2024) [1,2]. Here we present the results of the virological and epidemiological investigation of this case, including molecular characterisation of the virus and an assessment of the likelihood of onward transmission.

Case description and virological findings 

In mid-March 2026, an adult patient presented to the emergency department of our hospital, major tertiary referral centre in the Lombardy Region, Italy. They had experienced fever and cough since mid-January, accompanied by notable weight loss. They had returned from Senegal on the day of admission, having stayed there for more than 6 months [2]. The patient did not seek medical care or take any medication during their stay in West Africa. They recognised and self-monitored fever. Upon arrival, they were clinically stable, with an oxygen saturation of 97% on room air and a body temperature of 38.1°C.

 Laboratory findings showed anaemia, hyponatraemia and elevated lactate dehydrogenase (Table 1). A nasal-pharyngeal swab (NPS) tested negative for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), influenza A virus (IAV), influenza B virus and respiratory syncytial virus (RSV) (Table 2).

 A chest X-ray showed consolidation in the right middle and lower lung fields, and a small pleural effusion. A chest computer tomography (CT) scan revealed extensive consolidation of the left upper lobe, diffuse bilateral micronodules and a large right pleural effusion. An abdominal CT scan showed multiple hypodense lesions on the spleen and moderate ascites. 

As miliary tuberculosis was suspected, the patient was admitted to a single negative-pressure isolation room under airborne isolation precautions. Two days after admission, analysis of a sample from bronchoalveolar lavage (BAL) confirmed the presence of Mycobacterium tuberculosis. Anti-tuberculosis therapy comprising rifampicin, isoniazid, ethambutol and pyrazinamide was initiated. Further immunological evaluation revealed considerable cellular immunosuppression (Table 1).

The sample from BAL was tested using two commercial multiplex assays for respiratory virus detection, revealing a positive result for IAV. However, the H1pdm09 and H3 subtyping assays were negative (Table 2). According to the regional pandemic preparedness plan for influenza, all respiratory samples testing positive for IAV but negative for the seasonal subtypes should be sent immediately to a regional reference laboratory (RRL). 

There are three RRLs in the Lombardy region: the University of Milan, the Fondazione IRCCS Policlinico San Matteo and the ASST Fatebenefratelli-Sacco [3]. Further real-time RT-PCR testing at the University of Milan RRL confirmed the presence of IAV, with no detection of the H1pdm09 or H3 seasonal subtypes or avian A(H5N1) or A(H7N9). Given the suspicion of a zoonotic IAV infection, the regional authorities and the National Influenza Centre (NIC) were alerted at once, and oseltamivir therapy (75 mg twice daily) was initiated. 

On day 6 after admission, real-time RT-PCR was performed to detect avian IAV subtypes H5, H7 and H9, and a positive result was obtained for H9. On that day, nasal and throat swabs were collected from the patient, with only the throat swab testing positive for IAV. According to the national procedure [4,5], an aliquot of BAL sample was sent to the NIC where it was confirmed as an IAV subtype H9 (Table 2). The virus was isolated in Madin–Darby canine kidney (MDCK) cells (American Type Culture Collection (ATCC), CRL-2935) at both Fondazione IRCCS Policlinico San Matteo RRL and NIC.

(SNIP)

Discussion  

To our knowledge, this is the first reported human case of avian influenza A(H9N2) in Europe [2,17]. The detection of an unsubtypable IAV in the patient with severely weakened immune system prompted a thorough molecular investigation, including characterisation of the virus, which highlights the effectiveness of the diagnostic and surveillance system.

The regional public health authorities identified, tested and interviewed 13 contact persons. Nevertheless, contact tracing is challenging when airline companies and tour operators are involved. Several individuals could not be traced; however, all those who were successfully traced and tested, returned negative results.

The genetic similarity of the virus to previously detected strains in West Africa suggests that the patient may have been exposed to the virus during their time in the region, despite reporting no direct contact with animals. The presence of molecular markers associated with human receptor binding further highlights the zoonotic potential of A(H9N2) viruses. However, there is currently no evidence of human-to-human transmission.

Notably, the initial NPS was negative for IAV, potentially due to inadequate specimen collection or a low viral load in the upper respiratory tract at the time of sampling. In this patient with weakened immune system, the infection was initially detected in the lower respiratory tract, as evidenced by BAL positivity. Later NPS positivity, however, was associated with high quantification cycle (Cq) values and suggested the detection of residual viral RNA rather than active replication in the nasopharynx [18].

Conclusion 

The potential for prolonged replication in patients with weakened immune systems raises concerns about the emergence of escape variants, emphasising the need for continued vigilance. This case underlines the importance of considering non-seasonal influenza viruses in patients with compatible symptoms and relevant travel history and highlights the added value of genomic characterisation in the public health response.

While this hospital did an admirable job in diagnosing this patient - had this patient not had a concurrent severe lung infection, or had not been admitted to a modern hospital - H9n2 might easily have been missed. 

The patient had reportedly been ill in Senegal for at least a month, and while his route of exposure to H9N2 is unknown, his lack of contact with poultry or farm environments suggests at least the possibility of a community acquired infection. 

We'll never know, of course. 

Recent surveillance reports, however, have indicated that H9N2, along with H7 and H5 viruses, have been detected in both pigs and poultry in Senegal (see Influenza A Virus in Pigs in Senegal & Risk Assessment of AIV Emergence and Transmission to Humans).

While we comfort ourselves with our current low number of human novel flu detections, we are probably missing some number of cases.  

For more on this, you may wish to revisit UK Novel Flu Surveillance: Quantifying TTD, which suggests it might take weeks, and hundreds of cases, before our surveillance systems would detect low-level community spread of a novel flu virus.

And of course by that time, our options for containment would be limited.   

EID Journal: Highly Pathogenic Avian Influenza A(H5N1) Virus RNA in Bovine Semen, California, USA, 2024

 

#19,121

A little over two years ago HPAI H5N1 was detected in dairy cattle in Texas and Kansas, which prompted me to write A Brief History Of Influenza A In Cattle/Ruminants, where we looked at a number of past papers on both influenza and influenza-like illnesses in cattle and goats.

While relatively rare, both influenza and influenza-like illnesses in cattle and goats had been observed prior to 2024. In some cases, viruses were identified, while in other cases they were not.

Over the next twelve months - despite only limited testing - we saw hundreds of dairy herds across more than a dozen states infected with HPAI, along with sporadic spillovers into goats, alpacas, pigs, and even a sheep in the UK.

From the beginning, testing was restricted almost exclusively to lactating dairy cows - and for practical/logistical reasons that testing has often been indirect - via bulk tank testing of milk. 

• The USDA's Dairy Herd Status Program website hasn't been updated since Jan 8th, 2006 , but in its first year showed just 130 herds (out of an est. 36,000) from 21 states enrolled in the voluntary herd monitoring program.  

• Far more testing is done, however, via the National Milk Testing Strategy (NMTS) and individual state programs, which test pooled milk from many farms and is primarily used to identify which regions/states have the virus. 

Since the prevailing theory is that HPAI mainly affects lactating dairy cows - due to the virus's affinity to bovine mammary cells - very little testing of dry cows, heifers, calves, or beef cattle (bulls & steers) has been done. 

This, despite a March 2025 study (see Virology: Detection of Antibodies Against Influenza A Viruses in Cattle)  which reported that bulls and steers were just as likely to carry antibodies to (non-HPAI H5) IAV as cows and heifers (also see a more recent report on H1N1 in a European (male) Bison).

Admittedly, the number of non-lactating cattle far exceeds the number of dairy cattle in the U.S., and the logistics of testing and tracking individual (often asymptomatic) beeves would be daunting.  

But this represents a massive, largely untested, potential reservoir for the virus. 

All of which brings us to a new report in the EID Journal which finds (limited) evidence of H5N1 RNA in bovine semen, which could present an inadvertent (and largely hidden) way to spread the virus. 

In short, researchers tested 3 bulls on a California dairy farm during an outbreak of H5N1 (B3.13) in October of 2024. They found low-levels of H5N1 RNA in a semen sample from one bull (confirmed by PCR).  No infectious virus was isolated, possibly due to the weak sample.

While something less than a smoking gun, this is enough of a signal to warrant additional research. I've posted the link, and some highlights, from the report.  I'll have a bit more after the break.

Research Letter

Highly Pathogenic Avian Influenza A(H5N1) Virus RNA in Bovine Semen, California, USA, 2024

 

Ailam Lim1, Keith Poulsen1, Leonardo C. Caserta, Lizheng Guan, Eryn Opgenorth, Maxwell P. Beal, Amie J. Eisfeld, Yoshihiro Kawaoka, and Diego G. Diel

Abstract

Since March 2024, highly pathogenic avian influenza (HPAI) A(H5N1) virus has infected dairy cattle in the United States, prompting concern about novel transmission routes. During an outbreak in California, HPAI H5N1 RNA was detected in an asymptomatic bull’s semen. Although infectious virus was not isolated, semen-associated transmission risks and biosecurity practices remain a concern.

Since March 2024, detection of clade 2.3.4.4b highly pathogenic avian influenza (HPAI) A(H5N1) in US dairy cattle has raised concerns about the virus’s ability for cross-species transmission, adaptation to mammals, and novel transmission routes, including milk (1,2). Multiple pathogenic viruses are transmitted in bovine semen, and detection of HPAI in turkey semen has prompted questions about the potential role of HPAI transmission in bovine semen (3,4).

 Shedding of HPAI H5N1 in bovine semen could result in silent viral spread within herds and across geographic regions through artificial insemination. Although clinical HPAI disease has been reported in female calves and pregnant animals, reports of diseased bulls in dairy farms or beef cattle are lacking.

Many questions about the pathophysiology of HPAI H5N1 in US dairy herds remain unanswered, but movement of lactating cows is a recognized risk factor for interstate disease spread. In this diagnostic study, we sought evidence of HPAI H5N1 shed through semen in natural breeding bulls on an HPAI H5N1–affected dairy farm in California.

(SNIP)

Because of the limited semen volume available for analysis, we did not perform further confirmation testing at the national reference laboratory. We requested additional samples several months later for convalescent testing, but bull 1 had been culled from the herd. The significance of identifying HPAI H5N1 in bovine semen remains uncertain. The virus could have been actively shed in semen, or the ejaculate could have been contaminated during collection.

Although detecting RNA does not confirm the presence of infectious virus, this finding warrants further investigation into whether HPAI H5N1 can be shed in semen and raises questions about farm biosecurity amid the ongoing outbreak. High viral load in the environment during a herd outbreak was well documented (S. Lakdawala et al., unpub. data, https://www.biorxiv.org/content/10.1101/2025.07.31.666798v3; C. Stenkamp-Strahm et al., unpub. data, https://www.medrxiv.org/content/10.1101/2025.09.03.25335023v1). Good biosecurity measures are essential to prevent infections and, if infection occurs, slow disease spread on the farm.

In conclusion, further research and risk assessments are needed to determine tissue tropism of HPAI H5N1 in reproductive organs and whether naturally infected bulls shed virus in semen, and, if so, evaluate the risk for disease spread on dairy farms and with artificial insemination programs. Repetition and confirmation of these findings would have implications for natural breeding and biosecurity for artificial insemination collection centers, suggesting the need for increased caution in preventing silent intraherd spread.

(Continue . . . )

More than 2 years after the first discovery of H5N1 in dairy cows in Texas there remain far too many unanswered questions regarding the prevalence, and spread, of HPAI in American (and global) livestock. 

Over that time we've seen a number of studies suggesting that Influenza A in general (including HPAI H5), can and does infect livestock, including bovines, pigs, and goats. A few of many studies include:

Netherlands: NOS.NL Reports 5 Dairy Cows Have Now Tested Positive for H5N1

Transboundary & Emerg Inf: Serological Evidence of HPAI (H5N1) in Invasive Wild Pigs in Western Canada,

Preprint: Highly Pathogenic Avian Influenza H5 Virus Exposure in Goats and Sheep (in Pakistan).

Ten days ago, in Preprint: Bovine H5N1 Influenza Viruses Have Adapted to More Efficiently Use Receptors Abundant in Cattle, we saw new evidence that  H5N1 was undergoing active and robust viral adaptation in cattle.

And yet there appears to be little sense of urgency among farmers, or regulatory agencies. Surveillance remains largely passive, and in some regions of the world, practically non-existent.

While H5 may find a dead-end in cattle, we are tempting fate by ignoring the current trajectory, which suggests HPAI is increasing in - and adapting to - farmed mammals.  

Nature Med.: Rapid expansion of genotype D1.1 A(H5N1) influenza viruses in wild birds across North America during the 2024 migratory season

 
Abrupt Shift in H5N1 Genotypes in Wild Birds in US/Canada

#19,120

After its arrival in eastern Canada in late 2021, HPAI H5N1 spread quickly to the U.S. - reassorting with local LPAI avian flu viruses as it traveled - producing more than 100 new genotypes.  

While most of these reassortants were unremarkable, in early 2024 a B3.13 genotype emerged that was particularly well-suited for infecting dairy cattle. Six months later, another new genotype appeared in North American birds; D1.1. 

In short order (see chart above), this upstart D1.1 genotype went from zero detections to dominating in wild birds, spilling over into poultry farms, infecting poultry workers, and sometimes displaying signs of antiviral resistance (see Emerg. Microbes & Inf: Oseltamivir Resistant H5N1 (Genotype D1.1) found On 8 Canadian Poultry Farms).

Unlike the `bovine' strain - which typically produced mild (primarily conjunctival) symptoms in humans, D1.1 produced more serious illness in a number of cases.

• In November of 2024, we learned of a severe case involving a teenage girl in British Columbia (see NEJM: Critical Illness in an Adolescent with Influenza A(H5N1) Virus Infection)
• The following January the Louisiana Department of Health Announced the 1st U.S. H5N1-related Fatality
• And a Fatal case was reported in Mexico (see WHO DON Update On Mexico's Fatal H5N1 Infection) in April.

In addition to this increased virulence in humans, D1.1 further demonstrated its versatility by spilling over into cattle in at least 3 states (Nevada, Arizona & Wisconsin).  

Recent studies suggest it may also be better adapted to human physiology (see J.I.D.: Avian influenza virus A(H5N1) genotype D1.1 is better adapted to human nasal and airway organoids than genotype B3.13).

All of which brings us to a Brief Communications - published in Nature Medicine - which details the origins, and rapid spread, of D1.1 in North American birds.  I've only posted some brief excerpts, so follow the link to read it in its entirety. 

I'll return with more after the break.

Published: 15 April 2026

Rapid expansion of genotype D1.1 A(H5N1) influenza viruses in wild birds across North America during the 2024 migratory season

Walter N. Harrington, Anthony Signore, Lisa Kercher, Jolene A. Giacinti, Ahmed Kandeil, Christina A. Ahlstrom, Sarah Bevins, Beate Crossley, Karlie Eure, Thomas P. Fabrizio, Trushar Jeevan, Julianna Lenoch, Jacqueline M. Nolting, Daniel Rejmanek, David Stallknecht, Trent Bollinger, Evan J. Buck, Deborah Carter, Bradley S. Cohen, Krista E. Dilione, Jamie C. Feddersen, John Franks, Dayna Goldsmith, Cory J. Highway, …Richard J. Webby 

Nature Medicine (2026)Cite this article 

Abstract

In late 2021, high pathogenicity avian influenza A(H5N1) clade 2.3.4.4b viruses entered North America and reassorted rapidly with local avian influenza viruses. In September 2024, we detected a new reassortant later classified as genotype D1.1. Using active and passive avian influenza surveillance across Canada and the USA, we tracked the emergence and rapid spread of D1.1 viruses in wild birds during the 2024 fall migration.

Phylodynamic analysis showed that D1.1 viruses formed a monophyletic group and displaced earlier A(H5) genotypes across several flyways. Their expansion coincided with detections in other hosts, including 17 human cases, 4 of which were severe or fatal. None of the mammalian-adaptive markers detected in human cases were found in wild bird viruses, and candidate vaccine viruses retained antigenic cross-reactivity with D1.1 strains.

(SNIP)

Conclusion 

The emergence and rapid spread of HPAI A(H5N1) genotype D1.1 viruses across North American migratory flyways during the 2024 migration season represents a notable shift in clade 2.3.4.4b epidemiology. Several factors may explain the unusually efficient transmission of D1.1. Genomic analyses show that this genotype includes a distinct North American–derived neuraminidase (NA) segment, which may provide antigenic advantages because population immunity to this NA is probably lower than to NA segments circulating widely before 2024.

Although principal mammalian adaptation markers were absent in the wild bird samples, some D1.1 viruses carried several mutations with reported phenotypic effects (Supplementary Table 2), which may have contributed to their dominance. Increased densities of immunologically naive juvenile birds during southbound migration may have facilitated rapid viral amplification at staging areas14.

Environmental conditions also probably played some role in the dissemination of genotype D1.1 viruses. Persistent drought and altered habitat use at important breeding and staging areas may have led to increased mixing and density of wild birds, facilitating interspecies viral transmission15. These ecological factors, combined with the novelty of the genotype in wild populations, may have enabled the rapid geographic spread observed. 

Although D1.1 viruses circulating in wild birds currently lack principal mammalian adaptation markers, their sustained replication in mammalian hosts, such as US dairy cattle16, increases the evolutionary potential for zoonotic transmission. D1.1 viruses detected in poultry in British Columbia were found to carry the NA-H275Y mutation—a known marker for resistance to oseltamivir17, demonstrating that resistance and adaptation markers can emerge stochastically, even without selection pressure. 

(SNIP)

In conclusion, the emergence of the D1.1 viruses coincided temporally with the southward autumn migration of North American waterfowl, favoring their dissemination. Their spread highlights the need for an integrated One Health response18 that aligns wildlife surveillance, agricultural biosecurity and public health preparedness.

        (Continue . . . )

 

D1.1 likely remains a key player in North American birds, but the lack of genomic specificity in recent reporting makes it difficult to quantify how prevalent it remains in 2026.  The USDA's dashboard of wildbird detections does not specify genotype, and instead classifies viruses as:

• EA = Eurasian; AM = North American; the EA H5 (2.3.4.4) viruses are highly pathogenic to poultry.

• EA/AM: reassortant of H5 goose/Guangdong and North American wild bird lineage

A little over a year ago, in Nature: Lengthy Delays in H5N1 Genome Submissions to GISAID, we looked at some of the many obstacles in analysing surveillance data which are often missing crucial metadata (i.e. collection date, exact location, host-specific information, genotype, etc.).

While D1.1 has demonstrated remarkable ecological success - and a concerning ability to spillover into mammals - it doesn't appear to be ready for prime time. 

It does, however, remind us how abruptly the novel fluscape can change.  

Whatever critical zoonotic traits D1.1 (and other genotypes) currently lack could easily be rectified by some future untoward reassortment event.

But without substantially improved wildlife and agricultural surveillance under a One Health framework - and the rapid and open sharing of data - we'll likely never see it coming. 

Obstetrics & Gyn: Recovery of Pregnancy-Related Death Ratios After the Coronavirus Disease 2019 (COVID-19) Pandemic

CDC MMWR Sept 2020


#19,119

While we've known that pregnant women and their unborn offspring are often among the hardest hit during influenza pandemics (see 2009's Pregnancy & Flu: A Bad Combination), COVID's impact on pregnancy remains less well understood. 

Early in the pandemic we looked at a number of studies that found increased risks for pregnant women, including:

• Preprint: Impact of SARS-CoV-2 Variant on the Severity of Maternal Infection and Perinatal Outcomes

• AJOG: Disease Severity, Pregnancy Outcomes and Maternal Deaths With Patients With SARS-CoV-2 - Washington State

• MMWR: Two New Reports On Pregnancy & COVID-19

• PAHO Epi Alert: COVID-19 During Pregnancy - 13 August 2020

• Pregnancy & COVID-19: Still More Questions Than Answers

And a 2023 Meta-Analysis (Adverse maternal, fetal, and newborn outcomes among pregnant women with SARS-CoV-2 infection) published by the BMJ reported:

• ~3–4× higher risk of ICU admission
• ~5× higher need for critical care
• ~7–8× higher risk of maternal death (relative risk)
• Much higher risk of mechanical ventilation and pneumonia

But today we've a analysis, published in Obstetrics & Gynecology, that quantifies the impact of pregnancy-related deaths during the COVID pandemic (up 60% in 2021), and the fact that those deaths - while lower now - still remain elevated for some sociodemographic subgroups

The full report is very much worth reading.  I'll have more after the break.

Original Research

Recovery of Pregnancy-Related Death Ratios After the Coronavirus Disease 2019 (COVID-19) Pandemic

Disparities by Age, Race and Ethnicity, and Geography

MacCallum-Bridges, Colleen L. PhD; Daw, Jamie R. PhD; Admon, Lindsay K. MD, MSc

Obstetrics & Gynecology ():10.1097/AOG.0000000000006255, April 2, 2026. | DOI: 10.1097/AOG.0000000000006255

 Abstract

 

OBJECTIVE: 

To describe trends in pregnancy-related death ratios from 2018 to 2024, assess the contribution of coronavirus disease 2019 (COVID-19) to these trends, and evaluate whether pregnancy-related death ratios have recovered to prepandemic levels.

METHODS: 

We conducted an observational study that used vital statistics data to calculate the annual pregnancy-related death ratio (the number of pregnancy-related deaths per 100,000 live births) for female individuals aged 15–49 years between 2018 and 2024. We compared the pregnancy-related death ratios across prepandemic (2018–2019) pandemic (2020–2022), and postpandemic (2023–2024) periods; to assess the contribution of COVID-19, we calculated the pregnancy-related death ratio including and excluding COVID-associated deaths (ie, those with ICD-10 U07.1 listed as a cause). Pregnancy-related deaths were identified using International Statistical Classification of Diseases, Tenth Revision codes (A34, O00–O99), and the total pregnancy-related death ratio was decomposed into the early pregnancy-related death ratio (deaths during pregnancy or within 42 days after pregnancy) and the late pregnancy-related death ratio (deaths 43–365 days postpartum). We conducted subgroup analyses by maternal age, race and ethnicity, or geographic region.

RESULTS: 

From 2018 to 2024, there were 8,298 pregnancy-related deaths (32.3/100,000 live births). From the prepandemic period to the pandemic period, the early pregnancy-related death ratio increased by 7.5 deaths per 100,000 live births (95% CI, 6.1–8.8) and the late pregnancy-related death ratio increased by 3.7 deaths per 100,000 live births (95% CI, 2.7–4.6). Most of this increase (76% for the early pregnancy-related death ratio, 50% for the late pregnancy-related death ratio) was COVID-associated deaths. 

By 2023–2024, the early pregnancy-related death ratio had returned to prepandemic levels, but the late pregnancy-related death ratio remained elevated (1.4 additional deaths/100,000 live births; 95% CI, 0.4–2.4). Most subgroups experienced an increase in early and late pregnancy-related death ratios during the pandemic, but recovery varied.

Notably, both early and late pregnancy-related death ratios remained substantially elevated among non-Hispanic Black mothers in 2023–2024 compared with the prepandemic period (early pregnancy-related deaths increased by 7.0/100,000 live births [95% CI, 1.3–12.8]; late pregnancy-related deaths increased by 5.4 /100,000 live births [95% CI, 1.3–9.5]).

CONCLUSION: 

Pregnancy-related death ratios increased dramatically during the COVID-19 pandemic, and by 2023–2024, recovery differed by the timing of death relative to pregnancy and across sociodemographic subgroups. Additional efforts are needed to identify drivers of differential recovery from the COVID-19 pandemic and inform clinical and policy initiatives to reduce pregnancy-related deaths, improve maternal health, and promote health equity.

       (SNIP)

Pregnancy-related deaths increased dramatically during the COVID-19 pandemic, peaking in 2021 with 32.7 early pregnancy-related deaths per 100,000 live births and 13.1 late pregnancy-related deaths per 100,000 live births. Pandemic era increases in pregnancy-related deaths were largely explained by COVID-associated deaths, but by 2023–2024, recovery from the COVID-19 pandemic differed across sociodemographic subgroups.

Although some groups had returned to prepandemic levels, pregnancy-related death ratios remained elevated for other groups, including non-Hispanic Black mothers, a demographic that already faced substantially higher rates of pregnancy-related deaths before the COVID-19 pandemic. Additional research is needed to identify drivers of differential recovery from the COVID-19 pandemic and inform clinical, public health, and public policy initiatives to reduce pregnancy-related deaths and promote maternal health and health equity.

       (Continue  . . . ) 

As the following CDC chart illustrates, pregnancy related mortality rose sharply in 2020, peaked in 2021, and began dropping significantly in 2022.

It is worth noting that while available, uptake of the COVID vaccine was quite limited among expectant mothers in 2021, with the CDC reporting:

Data from the COVID-19-Associated Hospitalization Surveillance Network (COVID-NET) in 2021 indicate that approximately 97% of pregnant people hospitalized (either for illness or for labor and delivery) with confirmed SARS-CoV-2 infection were unvaccinated.

In late September 2021 the CDC Issued a HAN For Pregnant Women Urging COVID Vaccination, which helped to increase its uptake.  

It seems likely that the COVID vaccine, along with acquired immunity, better medical care options, and the arrival of less virulent Omicron variant in 2022, all contributed to this sharp decline. 

Today, according to the latest (March 2026) ACOG COVID-19 Vaccination Considerations for Obstetric–Gynecologic Care Practice Advisory update, only about 11% of pregnant women in 2026 have received a COVID-19 vaccine. 

However, vaccine uptake has waned, with a 25.7% year-to-date decrease in the 2025–2026 season compared with 2024–2025 (CDC 2025d). As of February 21, 2026, 11.1% of pregnant women overall have received a COVID-19 vaccine (CDC 2026). Clear, strong clinician recommendation remains one of the most influential factors in maternal vaccination acceptance and is essential to reducing preventable morbidity.

While many choose to view COVID as a non-threat today, it continues to cause significant morbidity and mortality (see WHO Statement: COVID-19 Still Causes Severe Disease & Renewed Vaccination Recommendations).

A few recent studies include:

Nature Comms: The Risk of Kidney Disease Increases Following SARS-CoV-2 Infection Compared to influenza

EID Journal: Thrombotic Events and Stroke in the Year After COVID-19 or Other Acute Respiratory Infection

Referral: JAMA - COVID-19 in Pregnancy Linked With Risk of Neurodevelopmental Disorders in Early Childhood

European Society of Cardiology: Major Consensus Statement Released on Long-Term Cardiovascular Impact of COVID Infection

Unfortunately, our growing societal laissez-faire attitude towards this virus means that some people are only going to learn this the hard way. 

The Lancet Microbe: Characterisation of Immune Responses Targeting HPAI A(H5) Viruses in Health-care Workers in the Netherlands

 

Credit CDC Antigenic Characterization

#19,118

Until relatively recently it was assumed that most of the world's population had little to no immunity against HPAI H5 viruses. In June of 2024 (see CDC A(H5N1) Bird Flu Response Update: Population Immunity to A(H5N1) clade 2.3.3.4b Viruses) the CDC reported: 

Data from this study suggest that there is extremely low to no population immunity to clade 2.3.4.4b A(H5N1) viruses in the United States.  Antibody levels remained low regardless of whether or not the participants had gotten a seasonal flu vaccination, meaning that seasonal flu vaccination did not produce antibodies to A(H5N1) viruses.

This means that there is little to no pre-existing immunity to this virus and most of the population would be susceptible to infection from this virus if it were to start infecting people easily and spreading from person-to-person. This finding is not unexpected because A(H5N1) viruses have not spread widely in people and are very different from current and recently circulating human seasonal influenza A viruses.

Since then, however, we've seen a number of studies which have reported serological evidence of limited immunity - particularly in older cohorts - against some strains of H5N1.  

Probably not enough to prevent infection, but perhaps enough to reduce the severity of infection. 

A few recent examples include:

Nature Comms: Pre-existing Cross-reactive Immunity to HPAI 2.3.4.4b A(H5N1) Virus in the United States

mBio: Low levels of influenza H5N1 HA and NA antibodies in the human population are boosted by seasonal H1N1 infection but not by H3N2 infection or influenza vaccination

Preprint: Cross-Reactive Human Antibody Responses to H5N1 Influenza Virus Neuraminidase are Shaped by Immune History

How much `real-world' protection any of this might offer is anyone's guess. But with a pandemic specific vaccine unlikely in the opening months of an outbreak, and growing concerns over our antiviral armamentarium, any immunological edge - no matter how small - would be welcome. 

While most of the studies we've seen to date have focused on the American Bovine B3.13 strain, today we've a study from the Netherlands which uses two European clade 2.3.4.4b H5N1 viruses along with an older Chinese (2005) clade 2.3.4 virus. 

The research summary reads:

Research in context

Evidence before this study

We searched PubMed from Jan 1, 1997, to July 1, 2025 in English using combinations of “influenza”, “heterosubtypic”, “immunity”, “cross-immunity”, “baseline”, “population”, “humans”, “H5N1”, “avian influenza”, “clade 2.3.4.4b”, “humoral”, “antibodies”, “hemagglutinin”, “neuraminidase”, “cellular”, and “T-cells”. Previous reports on immunity to A(H5) influenza viruses on a population level are fragmented, because most examined binding or functional antibodies exclusively, and few examined T-cell responses.

Added value of this study

We profiled immune responses targeting A(H5) influenza viruses in a presumed unexposed cross-sectional cohort of health-care workers (n=107) by combining the measurement of antibody binding, haemagglutinin inhibition, antibody-dependent cellular cytotoxicity (ADCC), neuraminidase (NA) inhibition, and T-cell responses. A(H5) antibodies binding the haemagglutinin stalk, and NA inhibition-mediating and ADCC-mediating antibodies were common. Cross-reactive T cells targeting the haemagglutinin and NA of A(H5) influenza viruses were detected. Our rational selection of antigens from both seasonal and avian influenza viruses enabled a direct comparison of multiple effector mechanisms, establishing the first integrated baseline map of humoral and cellular responses against A(H5) influenza viruses.

 

Implications of all the available evidence

Our findings, in combination with existing research, show that partial immunity to A(H5) influenza viruses is widespread. Repeated exposure to seasonal influenza viruses likely led to the development of a cross-reactive antibody and T-cell repertoire recognising A(H5) influenza viruses. Although the protective value of cross-reactive immune responses remains speculative, evidence from animal model systems suggests that these responses might confer partial protection.

The full study, while fairly technical, is well worth reading. I've posted the abstract below.  I'll have a bit more after the break.

Characterisation of immune responses targeting highly pathogenic avian influenza A(H5) viruses in health-care workers in the Netherlands: an observational, cross-sectional analysis

Mark A Power, MSca,∗ ∙ Willemijn F Rijnink, MSca,∗ ∙ Widia Soochit, PhDa ∙ Lennert Gommers, BSca ∙ Anne van der Linden, BSca ∙ Felicity Chandler, BSca ∙ Femke Volker, BSca ∙ Theo M Bestebroer, BSca ∙ Babs E Verstrepen, PhDa ∙ Alba Grifoni, PhDb ∙ Ngoc H Tan, MPharma ∙ Susanne Bogers, MSca ∙ Gijsbert P van Nierop, PhDa ∙ Prof Alessandro Sette, PhDb,c ∙ Prof Marion P G Koopmans, PhDa ∙ Corine H Geurts van Kessel, PhDa,† ∙ Reina S Sikkema, PhDa,† ∙ Mathilde Richard, PhDa,‡ ∙ Rory D de Vries, PhDa,‡ r.d.devries@erasmusmc.nl Show less

 

Published April 13, 2026

DOI: 10.1016/j.lanmic.2026.101367 External Link

 

Download PDF

 

Summary

Background

Highly pathogenic avian influenza A(H5) viruses pose a pandemic threat, with a history of mammalian adaptation and zoonotic spillovers into humans. We aimed to determine whether pre-existing cross-reactive immune responses to A(H5) clade 2.3.4.4b influenza viruses detected between 2020 and 2024 are present in the general population.

Methods

We conducted an observational, cross-sectional study within the prospective Surveillance of Respiratory Viruses in Healthcare and Animal Workers in the Netherlands (SENTINEL) cohort, in which we analysed a subset of health-care workers aged 18 years or older who provided blood samples at a periodic study visit in August or September, 2024. Blood samples were analysed for influenza A(H5)-specific antibody binding, haemagglutination inhibition, Fc-effector functions, neuraminidase (NA) inhibition, and T-cell responses.

Findings

We included 107 health-care workers. Participants’ median age was 50·0 years (IQR 40·0–58·0); 77 (72%) health-care workers were female, 29 (27%) were male, and one (1%) did not report their biological sex. Virus-specific antibodies were measured in 106 serum samples.

Low-level binding antibodies directed against the A(H5) haemagglutinin (HA) head were detected in up to 28 individuals (depending on the antigen), but without haemagglutination inhibition activity.

Nevertheless, we detected A(H5)-reactive antibodies with Fc-effector functions in all participants. Additionally, we observed high levels of antibodies with NA inhibition activity (geometric mean titre 208 [95% CI 153–284]) in up to 97% of the health-care workers against avian N1, and T-cell responses against HA and NA from A(H5) influenza viruses in 43–69% (46–74 of 107) health-care workers. A(H5)-specific responses correlated with immune responses targeting A(H1N1).

Interpretation

Together, our findings suggest that partial cross-reactive immunity to A(H5) influenza viruses exists in humans, likely induced by previous exposures to seasonal influenza viruses. This partial cross-reactive immunity might play an important role during future outbreaks, potentially by blunting disease severity. Characterising pre-existing baseline immunity is crucial for accurate pandemic risk assessment and preparedness planning.

       (Continue . . . )

While some of the headlines overnight in the European press (see Likely no deadly bird flu pandemic: Immunity against regular flu also works on H5N1) overstate these findings, any degree of immunity to H5N1 beats no immunity at all. 

Since this study utilized a convenience sample of healthcare workers, it did not include children or adolescents (< 20), or the elderly (> 70). 

But they did report:  In an age-stratified analysis, antibody levels against A(H5) influenza virus HA0 antigens appeared higher in health-care workers older than 60 years.

This increased immunity is likely the consequence of a lifetime of exposure to influenza viruses, but the majority (61%) of the test cohort also reported receipt of the seasonal flu vaccine over the past 4 years

Since we've seen other studies showing some limited protection against H5N1 (in ferrets) from the seasonal flu vaccine, it is possible that prior flu vaccination contributed to - or amplified - some of the preexisting immunity reported in this study.

While I find plenty of good reasons to get the flu shot each year, perhaps for some, that will provide an added incentive to get one this fall.  

At least, one can hope.