(Credit: © ImageFlow - stock.adobe.com)
America’s Power And Water Systems Have Security Gaps. So Do Our Emergency Plans.
In A Nutshell
- The short answer: Research suggests neither the country nor most households are fully ready for a long, deliberate attack on critical systems, though officials so far blame this week’s outages on technical failures rather than hackers.
- The evidence: Federal watchdog reports, FEMA’s national household surveys and a nationwide study of power outages point to open security gaps in power and water systems, thin home preparations and frequent long blackouts that happen with no attacker involved at all.
- What you can do: Studies of real emergencies find that people tend to cooperate rather than panic, so a simple plan, a few days of supplies and some emergency savings may matter more than knowing who caused the next outage.
Most drivers find out whether they keep jumper cables in the trunk on the night the battery dies in an empty parking lot. Nobody checks on a sunny afternoon.
For many Americans, the past week has felt like that night. On Nov. 12, Microsoft 365 began failing, and problems lingered for about a day. Within an hour, users in 74 countries and territories had reported trouble. The next afternoon, the Federal Aviation Administration halted departures nationwide after its air traffic communications network broke down. The trouble started around 3:30 p.m. Eastern, and flights didn’t resume until 8:15 p.m.
Other reports of network outages and glitches across banks, technology companies, and even the NSA have sparked fears that something bigger is going on. Online, people keep asking the same thing: is this what a major cyberattack looks like?
So far, officials say no. The FAA says that, based on the evidence available, it has “no indication that this was a cyberattack.” It says aging equipment became unstable when its network was flooded with connection requests. Early findings pointed to two sites, including one where its network contractor had been putting in a new router. Microsoft blamed “a large number of servers being unexpectedly unavailable.” In their latest statements, neither had announced a final root cause.
The scare still raises a fair question, and research can help answer it: if a real attack came, could the country and the people in it get through it?
How Real Is the Threat to America’s Critical Systems?
Federal agencies say foreign hackers have already gotten inside some networks.
In February 2024, the Cybersecurity and Infrastructure Security Agency (CISA), the NSA and the FBI, joined by agencies from Australia, Canada, the United Kingdom and New Zealand, issued a warning about a Chinese state-sponsored group called Volt Typhoon. According to the agencies, the group was quietly getting into U.S. computer networks so it could disrupt or damage critical infrastructure if a serious crisis or conflict with the United States broke out. Coverage of the advisory reported that the hackers had been inside some networks for at least five years. At the time, a CISA official told reporters the U.S. had “not yet seen these actors take steps to disrupt the functionality of critical infrastructure.”
To picture what a successful attack on the power grid might do, the best guide may be a disaster that had no hacker in it. A peer-reviewed analysis in Energy Research & Social Science looked back at the February 2021 Texas freeze. At its peak, more than 4.5 million customers (over 10 million people) lost electricity, and some went days without it. The failures spread to drinking water treatment and medical care. The researchers estimate losses at $130 billion in Texas alone. When power goes out, the damage looks much the same whether ice or code caused it.
Is the Nation Prepared for a Cyberattack on Its Power and Water?
Government watchdogs say important gaps are still open.
In a March 2021 report, the Government Accountability Office (GAO) found growing cyber risk in the local distribution systems that bring electricity to homes and businesses. More of the equipment running those systems can be reached remotely or is linked to companies’ office networks, which gives intruders more ways in. GAO said how much damage such attacks could cause is not well understood. It also found that the Department of Energy’s plans fell short on these local systems, partly because officials had put power plants and high-voltage transmission first. GAO still lists its recommendation as open and says it was continuing to monitor progress as of March 2026.
Water may be even less protected. In 2024, GAO warned that hackers who got into drinking water systems could leave tap water contaminated with harmful levels of bacteria or chemicals. GAO also noted a reversal at the Environmental Protection Agency. In March 2023, the EPA decided that existing law already required drinking water systems to assess their cybersecurity. After legal challenges, it dropped that requirement seven months later. In testimony this past May, GAO said the sector includes close to 170,000 water and wastewater systems. It said the EPA has found that wastewater systems and some drinking water systems face no requirement to assess their cybersecurity risks, and that the agency’s legal power to change that is significantly limited.
A spot check by the EPA’s own inspector general shows why that matters. Its passive scan of 1,062 drinking water systems, which together serve more than 193 million people, found 97 systems serving about 26.6 million people with critical or high-risk vulnerabilities. The scan covered only large systems serving 50,000 or more people, and the inspector general did not report that any of the weaknesses had been exploited.
The broader backlog is large. GAO has made more than 4,400 cybersecurity recommendations to federal agencies since 2010, and more than 730 had not been fully carried out as of February 2026. The agency has treated protecting critical infrastructure from cyber threats as a “high-risk” area since 2003.
Past warnings don’t always turn into fixes. According to the Texas analysis, federal regulators advised the state to winterize its power plants after a 2011 freeze caused smaller outages. Those recommendations were only advisory, and at least three dozen plants that failed in 2011 failed again in 2021. After this week’s ground stop, the FAA said it had changed a setting on about 60 routers like the one involved. It also said it would rethink doing routine equipment replacements during the day instead of overnight.
The grid also struggles with ordinary problems. A 2023 study in Nature Communications went through outage records for 2,447 U.S. counties, home to about three-quarters of Americans, from 2018 to 2020. It counted 17,484 outages lasting eight hours or more. Each one meant at least a small share of a county’s customers (0.1%) went without power that long, so many hit only part of a county. The authors consider eight hours long enough to carry potential health risks.
Over 70% of the counties had at least one such outage. Most happened on days with extreme weather. On nearly 40% of the county-days with a long outage, none of the weather events the team tracked had occurred, and the authors say those outages were likely due to technical problems such as equipment failure or transmission delays. The data left out many small rural utilities, so the true count may be higher.
A country that already sees thousands of long outages, even partial ones, with no enemy behind them has limited room to absorb one caused on purpose.
Are Americans Prepared for a Cyberattack at Home?
National surveys suggest most people are not, especially for cyber threats.
FEMA’s 2022 National Household Survey, based on 7,145 responses, included a set of questions about cyberattacks. Among people asked about that threat, 65% registered as seeing it as a risk by FEMA’s measure. Even so, the agency reported that “two-thirds of people felt unprepared for one.” In that group, only 19% reported making an emergency plan, and 9% reported putting together or updating supplies.
General disaster readiness is somewhat better. In FEMA’s 2023 survey of 7,604 adults, 51% believed they were prepared for a disaster, up from 42% in 2017. Fewer had taken the specific steps that matter when phones, cards and power fail: 48% had put together or updated supplies, 37% had made a plan, and only 18% had tested a family communication plan. People facing economic hardship were less likely to take the steps that cost money, like stocking supplies (41% versus 50%).
The gaps were wider for adults 60 and older. Only 28% had made a plan, compared with 41% of younger adults, and 10% had tested a family communication plan, compared with 21%. One finding in the same survey was more encouraging. Among people who had read, seen or heard preparedness information, 92% took at least one basic step, compared with 38% of those who hadn’t.
Money is part of the problem. In the Federal Reserve’s latest household survey, 63% of adults said they could cover a surprise $400 expense with cash, savings or a credit card they’d pay off right away. Twelve percent said they couldn’t pay it by any means.
For a medically fragile group, the risk is more immediate. A CDC-led analysis estimated that, as of 2012, about 685,000 Americans lived at home while depending on powered oxygen concentrators or ventilators. The authors say that figure is likely an undercount. The rate climbed steeply with age, to roughly 2,185 per 100,000 people 65 and older, and was highest in the South and West.
The Nature Communications outage study found that these risks overlap in some places. Clusters of counties in Arkansas, Louisiana and Michigan had both frequent long outages and high social vulnerability or heavy use of home medical equipment. The most socially vulnerable counties had a median of three long outages a year, compared with one in the least vulnerable. Nationally, counties with the heaviest use of home medical equipment actually saw somewhat fewer long outages than others.
Picture a family on a cold evening when the card reader at the gas station goes dark, the phone can’t reach anyone, and the pharmacy can’t fill a prescription. Whether that evening is an inconvenience or an emergency depends mostly on what they arranged weeks earlier.
How Do People Hold Up When Systems Fail?
Better than many fear, research suggests, though the fear itself takes a toll.
Fear is real, even without physical damage. In three online studies of Israeli adults conducted between 2013 and 2016, researchers measured reactions to cyberterrorism, mostly by showing people simulated news clips of attacks. People who watched the clips reported more anxiety than those who saw no attack. That held even when the staged attack only cost people data or money, though stolen money alone didn’t significantly raise people’s sense of personal threat. A heightened sense of threat, in turn, went along with more support for internet surveillance, government regulation and forceful military retaliation.
One finding cut the other way: viewers’ confidence in the government’s ability to protect infrastructure didn’t drop. The authors suggest the effects could be stronger in countries like the U.S., where attackers are often harder to identify. A week of unexplained outages and rumors may produce some of the same strain, even when no hacker is involved.
How people behave tends to be steadier than how they feel. A 2019 review in Frontiers in Public Health pulled together research on how the public acts in emergencies. The authors concluded that “panic (over-reaction) is rare in crowds in emergencies,” and that survivors commonly support one another. Interviews with 90 survivors and 56 witnesses of the 2005 London bombings found that most people were among strangers, yet helping was common.
A survey of 1,240 survivors of Chile’s 2010 earthquake and tsunami found that people who felt a shared bond with others affected were more likely to give emotional support and, indirectly, to join practical efforts that helped the whole community. The authors also note that not everyone helps, and that groups prizing individual self-reliance have been linked to lower participation in coordinated hurricane evacuations. Most of this research covers sudden physical disasters, not multiday digital outages, so it may not translate perfectly.
The same review points to something officials control: communication. In a field exercise, an online experiment and a field experiment involving emergency decontamination, people were less anxious, cooperated more and moved through faster when responders explained what they were doing and gave regular updates.
Trust can’t be taken for granted. In the Israeli experiments, about a third of respondents said they wouldn’t simply follow official word that tap water was safe after a cyberattack on the water system, preferring to take extra precautions of their own. Texas offers a cautionary example. In testimony to a House committee in March 2021, Varun Rai, a co-author of the blackout analysis, described “an absolute lack of coordinated, consistent, and timely emergency communication to the people of Texas.”
Taken together, the evidence suggests the bigger risk is not mass panic but a slow pileup of small shortages: no savings cushion, no plan for reaching family, no backup for medical equipment, and no clear word from officials about what is happening.
That brings things back to the parking lot. Jumper cables work the same whether the battery died from cold weather or because someone left the headlights on. The research suggests outages work the same way. Whether the cause is a hacker, an ice storm or a contractor’s router, the households that do fine are usually the ones that checked the trunk ahead of time. For most Americans, this week was that chance: a warning that cost some missed flights and email, not a crisis.
Paper Notes
Limitations
Gross, Canetti and Vashdi (2017), Journal of Cybersecurity. The work consists of three online studies of Israeli adults conducted between 2013 and 2016. The abstract lists 1,124, 909 and 592 participants. The methods describe Study 3 as a two-wave panel of 522 people, and the results tables analyze 1,027, 907 and 522 respondents. Studies 1 and 2 were randomized experiments using simulated news-style video clips; Study 3 surveyed the same panel before and after a real, announced hacktivist campaign. Reactions to staged videos may differ from reactions to a real attack. The authors note that their online samples may skew younger and more tech-savvy, that the role of media coverage remains unstudied, and that effects may differ in countries where attackers are harder to identify than they were in these Israeli scenarios. Viewing the cyberterrorism clips did not reduce confidence in the government’s ability to protect infrastructure or data. When the only harm was financial loss (Study 2), threat perception was not statistically different from the control group, though anxiety was.
Drury and colleagues (2019), Frontiers in Public Health. This is a policy and practice review rather than a new study, drawing on case studies, interviews and surveys from emergencies such as the 2005 London bombings and the 2010 Chile earthquake and tsunami, along with decontamination exercises and experiments. The authors note that not all emergencies are marked by mutual support and not everyone helps, and that norms of individual self-reliance have been linked to lower participation in coordinated hurricane evacuations. The Chile survey was cross-sectional, and identification with fellow survivors predicted practical community support only indirectly. Most of the cited evidence comes from the United Kingdom and Chile and from sudden physical disasters rather than prolonged infrastructure outages.
Do and colleagues (2023), Nature Communications. This observational analysis covered only 2018 to 2020, too short to assess long-term trends. Its data source, PowerOutage.us, often lacked small rural utilities, leaving 563 counties (many in the Midwest and Mountain West) without reliable data. A county-level 8+ hour outage was counted when at least 0.1% of a county’s customers were without power for eight or more hours, so many counted outages affected only part of a county, and the customers without power could change during an outage. County-level analysis could mask local differences. Weather events could not be causally linked to outages because cause-specific outage data were unavailable. The medical equipment data covered only Medicare recipients. Nationally, counties in the highest quartile of Medicare home medical equipment use had fewer long outages (median 1 versus 2) than other counties, even though some high-use counties clustered with high outage counts. The paper reports the share of U.S. counties covered as 77.9% in one section and 78.9% in another.
Molinari and colleagues (2017), Journal of Public Health Management and Practice. This retrospective analysis used 2012 insurance claims from privately insured people and extrapolated the prevalence to the national population, which the authors say may bias the estimate, likely toward an undercount. Only ventilator and oxygen concentrator users were included; dialysis patients were excluded. The data are now more than a decade old.
Busby and colleagues (2021), Energy Research & Social Science. This is a retrospective case analysis of a single event, the February 2021 Texas freeze, written within months of it. Its $130 billion economic loss figure is an estimate drawn from outside sources, and some facts (such as the full list of plants that failed in 2021) were not yet public when it was written. Several co-authors were participants in or advisers to the Texas energy market (see disclosures).
FEMA National Household Surveys (2022 and 2023). Both are self-reported national surveys (7,145 responses in 2022; 7,604 in 2023) conducted by phone and web and weighted to reflect the general public. Self-reported preparedness may overstate or understate actual supplies and plans. The cyberattack findings come from the subset of 2022 respondents asked hazard-specific questions, and the summary document does not report that subsample’s size.
Federal Reserve SHED (2025 report). The survey asks about a hypothetical $400 expense; it measures financial cushion, not disaster preparedness directly.
GAO reports and CISA advisory. These are government assessments, not peer-reviewed studies. GAO itself states that the scale of potential impacts from attacks on grid distribution systems is not well understood.
EPA Office of Inspector General scan (Report 25-N-0004). The scan was a passive, outside-in snapshot taken on October 8, 2024, covering only drinking water systems that serve 50,000 or more people. The OIG page does not describe its methods or tools, and it does not report that any vulnerability was actually exploited. Conditions may have changed since the scan.
Funding and Disclosures
Do and colleagues (2023) report funding from the National Institute on Aging (RF1AG071024) and the National Institute of Environmental Health Sciences (P30ES009089 and 2T32ES007322). Drury and colleagues (2019) list funding from the Economic and Social Research Council (RES-000-23-0446), the Leverhulme Trust (F/00 230/AO), Centro de Medición Mide UC at Pontificia Universidad Católica de Chile and the Interdisciplinary Center for Social Conflict and Cohesion Studies (CONICYT/FONDAP/15130009), PhD studentships funded by Public Health England and a studentship funded by the Turkish Ministry of Education; the authors declare no commercial or financial relationships that could be construed as a potential conflict of interest, and two authors were affiliated with Public Health England. Molinari and colleagues (2017) state no funding source and declare no conflicts of interest. Gross, Canetti and Vashdi (2017) report grants to Canetti from the National Institute of Mental Health (R01 MH073687), the Israel Science Foundation (594/15) and the US-Israel Binational Science Foundation (2009460), and to Gross from the Israel Science Foundation (156/13); each study received university IRB approval. No competing-interest statement was found in the article. Busby and colleagues (2021) state the work was not directly sponsored by any agency or entity; Shidore’s time was supported by the UT Energy Institute, and open access was supported by the LBJ School Office of Research, the UT Austin Energy Institute and the Webber Energy Group. Their disclosure states that Rhodes and Webber consult for companies operating in the Texas electricity market through their firm IdeaSmiths LLC, that Smith consults for companies operating in the Texas electricity market through her employer, AB Power Advisors, that Webber is also affiliated with ENGIE SA, and that Gilbert is an employee of the Nuclear Innovation Alliance; the authors declare no known competing financial interests that could have appeared to influence the work. Do and colleagues (2023) declare no competing interests. FEMA, the Federal Reserve, GAO and CISA documents are government publications.
Publication Details
Gross, M.L., Canetti, D., and Vashdi, D.R. “Cyberterrorism: Its effects on psychological well-being, public confidence and political attitudes.” Journal of Cybersecurity, volume 3, issue 1, pages 49–58. Advance access February 15, 2017. DOI: 10.1093/cybsec/tyw018. Affiliation: School of Political Science, University of Haifa, Israel (all authors).
Drury, J., Carter, H., Cocking, C., Ntontis, E., Tekin Guven, S., and Amlôt, R. “Facilitating Collective Psychosocial Resilience in the Public in Emergencies: Twelve Recommendations Based on the Social Identity Approach.” Frontiers in Public Health, volume 7, article 141. Published June 4, 2019. DOI: 10.3389/fpubh.2019.00141. Affiliations: University of Sussex (Drury, Ntontis, Tekin Guven); Public Health England (Carter, Amlôt); University of Brighton (Cocking); Canterbury Christ Church University (Ntontis).
Do, V., McBrien, H., Flores, N.M., Northrop, A.J., Schlegelmilch, J., Kiang, M.V., and Casey, J.A. “Spatiotemporal distribution of power outages with climate events and social vulnerability in the USA.” Nature Communications, volume 14, article 2470. Received January 14, 2022; accepted April 14, 2023. DOI: 10.1038/s41467-023-38084-6. Affiliations: Department of Environmental Health Sciences, Columbia University Mailman School of Public Health (Do, McBrien, Flores, Casey); Vagelos College of Physicians and Surgeons, Columbia University (Northrop); National Center for Disaster Preparedness, Columbia Climate School (Schlegelmilch); Stanford University School of Medicine (Kiang); University of Washington School of Public Health (Casey).
Molinari, N.A.M., Chen, B., Krishna, N., and Morris, T. Study of at-home electricity-dependent durable medical equipment users (article title to be confirmed against the journal page). Journal of Public Health Management and Practice, 2017, volume 23, issue 2, pages 152–159. DOI: 10.1097/PHH.0000000000000345. Affiliation: Centers for Disease Control and Prevention, Atlanta, Georgia.
Busby, J.W., Baker, K., Bazilian, M.D., Gilbert, A.Q., Grubert, E., Rai, V., Rhodes, J.D., Shidore, S., Smith, C.A., and Webber, M.E. “Cascading risks: Understanding the 2021 winter blackout in Texas.” Energy Research & Social Science, volume 77, article 102106. Available online June 2, 2021. DOI: 10.1016/j.erss.2021.102106. Affiliations: University of Texas at Austin (Busby, Rai, Rhodes, Shidore, Webber); University of Colorado Boulder (Baker); Payne Institute, Colorado School of Mines (Bazilian, Gilbert); Georgia Institute of Technology (Grubert); AB Power Advisors, LLC (Smith); ENGIE SA (Webber).
Federal Emergency Management Agency. “2022 National Household Survey on Disaster Preparedness” public summary (fielded February 10 to April 20, 2022; 7,145 responses) and “2023 National Household Survey on Disaster Preparedness: Key Findings” (7,604 responses). Individual and Community Preparedness Division.
Board of Governors of the Federal Reserve System. “Economic Well-Being of U.S. Households in 2025.” Released May 13, 2026; survey fielded October 17 to 28, 2025.
U.S. Government Accountability Office. “Electricity Grid Cybersecurity: DOE Needs to Ensure Its Plans Fully Address Risks to Distribution Systems.” GAO-21-81, March 18, 2021. Recommendation status: open as of March 2026.
U.S. Government Accountability Office. “Critical Infrastructure Protection: EPA Urgently Needs a Strategy to Address Cybersecurity Risks to Water and Wastewater Systems.” GAO-24-106744, August 1, 2024. Follow-up testimony: “Critical Infrastructure Protection: Actions Needed to Address Persistent Cybersecurity Threats to the Water and Wastewater Sector,” GAO-26-109159, published May 21, 2026.
Rai, V. Written testimony before the U.S. House of Representatives Committee on Science, Space, and Technology, hearing “Lessons Learned from the Texas Blackouts: Research Needs for a Secure and Resilient Grid,” March 18, 2021. Affiliation: University of Texas at Austin (Energy Institute; LBJ School of Public Affairs). Testimony PDF.
U.S. Environmental Protection Agency, Office of Inspector General. “Management Implication Report: Cybersecurity Concerns Related to Drinking Water Systems.” Report No. 25-N-0004. Scan results dated October 8, 2024.
Cybersecurity and Infrastructure Security Agency, National Security Agency and Federal Bureau of Investigation, with partner agencies. Joint advisory on PRC state-sponsored Volt Typhoon activity, February 7, 2024.







