When Everything Is Breaking, What Do We Believe?

From cyberattacks and technology outages to AI-generated deception and flying jellyfish, information overload is becoming a business risk.

By Jennifer Gilligan | President, IntegraMSP | Nov. 17, 2026

At some point in the past few days, I found myself reading about flying jellyfish. Not the kind you find in an aquarium, but objects supposedly drifting through the sky. Social media offered videos, reports of sightings and even a claim that someone had found one dead in a backyard. I could not independently verify those claims, but after the week we have had, my immediate reaction was less skeptical than I would have expected: Flying jellyfish? Well, of course.

Then I caught myself. I lead a technology company, spend much of my time evaluating operational and security risks, and have a background in journalism. Evidence and verification are supposed to be second nature to me. Yet the volume of strange, consequential, and sometimes contradictory information had started to change what felt surprising. I did not suddenly believe in airborne marine life. I was simply exhausted by the possibility that practically anything could be the next headline.

That experience raises a larger question. When a constant stream of real disruptions, misleading claims and synthetic media overwhelms our ability to sort fact from fiction, what happens to our judgment? More pointedly, what happens when the people navigating that noise are also responsible for protecting employees, customers and businesses?

A week of real disruptions and manufactured emergencies

The past week provided no shortage of legitimate reasons to pay attention. Microsoft 365 experienced significant service disruptions, affecting tools many organizations rely on for routine communication and collaboration. On Nov. 13, a telecommunications problem prompted the Federal Aviation Administration to issue a nationwide ground stop that lasted more than four hours. Then, on Nov. 16, United Airlines experienced a separate technology outage that delayed departing flights. Banking customers encountered technical problems of their own. That same day, security researchers disclosed a supply-chain compromise involving Optimizely, a website experimentation platform used by major brands. Researchers found malicious JavaScript associated with 178 organizations, including recognizable news, financial, and technology sites. The code was designed to steal information, alter cryptocurrency transactions and, in some circumstances, target vulnerable mobile devices.

These were genuine events, although they were not all cyberattacks. The Optimizely incident involved malicious code delivered through a trusted third-party service. The Microsoft disruption, FAA ground stop, and United Airlines outage were reported as technical problems, not hostile intrusions. The FAA said unstable legacy networking equipment contributed to Friday’s failure; United said there was no indication its Monday outage was cybersecurity-related. There is no established link between the two aviation disruptions. The distinction matters. A business may experience an outage, data exposure, or financial loss regardless of the cause, but the response should be based on what actually happened rather than on a generalized sense that everything is under attack.

Meanwhile, viral social media posts predicted a 90-day nationwide blackout beginning Nov. 17 and tied it to a planned federal emergency alert. The Emergency Alert System test was real: FEMA and the Federal Communications Commission had scheduled it for radio and television. The blackout claim was unsupported. PolitiFact found that a purported presidential announcement circulated in support of the rumor had been manipulated from unrelated footage. Genuine outages and a routine government test were being assembled into a story the evidence did not support.

To someone encountering these items in rapid succession, the difference can disappear. A security advisory, a technical outage, a government announcement, and an invented catastrophe can all arrive with the same urgent headline treatment. The events are not necessarily connected, but the experience of consuming them can make them feel connected.

When news becomes noise

There was a time when people generally made a deliberate choice to watch a broadcast, read a newspaper or visit a news website. Today, many encounter news incidentally, inserted between family photos, advertising, entertainment and commentary. Social platforms can distribute excellent original journalism, but their recommendation systems are generally designed to keep people engaged, not to apply a newsroom’s verification standards. A carefully sourced report may appear directly beside satire, conjecture, or a manufactured video, each competing for attention in much the same visual format.

The numbers help explain why the experience can be so disorienting. Pew Research Center reported in February that 52% of U.S. adults felt worn out by the amount of news, and 60% had reduced their overall news consumption at some point. A separate Pew survey, published in October, found that 59% of U.S. adults get news from social media at least sometimes. In the Reuters Institute’s 2026 Digital News Report, only 25% of U.S. respondents said they trust most news most of the time. These measures come from different surveys; they are not parts of one population breakdown, but together they illustrate the tension between access, exhaustion, and confidence.

The credibility squeeze | U.S. adults
52%
News exhaustion
Worn out by the amount of news
60%
Pulled back
Have reduced overall news consumption at some point
59%
Social-media news
Get news on social media at least sometimes
25%
News trust
Trust most news most of the time

Sources: Pew Research Center (February and October 2026); Reuters Institute (2026). Separate surveys, different questions.

This does not mean the public has uniformly abandoned traditional reporting. Pew also found that 66% of American adults use news websites or apps for news at least sometimes. The important change is that reporting now reaches people through an increasingly complicated mix of direct and algorithmically selected channels, often without the context that helps them evaluate what they are seeing.

Where Americans get digital news- U.S. adults who get news from each source often or sometimes; respondents could select more than one.

Search

70%

News sites/apps

66%

Social media

59%

Email newsletters

34%

AI chatbots

15%

Source: Pew Research Center, survey conducted Aug. 24–30, 2026, published Oct. 19, 2026.

Faced with relentless warnings, people can react in opposite ways. Some become hypervigilant, interpreting each new development as evidence of an approaching crisis. Others disengage because the stream of emergencies has become impossible to process. Neither response is a reliable way to assess an actual threat. And although news fatigue does not automatically cause belief in conspiracy theories, an environment of exhaustion and uncertainty creates opportunities for unverified explanations to travel quickly.

AI has complicated the meaning of evidence

Artificial intelligence makes that challenge more serious. Synthetic images, video, and audio can now closely resemble authentic recordings, including familiar voices and recognizable public figures. The old habit of inspecting a strange video for obvious visual errors is increasingly inadequate. A manipulated clip may give an otherwise baseless claim the appearance of proof, while genuine footage can be dismissed as fake simply because the viewer is skeptical of its source.

In a study published in Communications of the ACM, 1,276 participants were asked to classify authentic and synthetic media across image, audio, and video formats. Mean detection accuracy was 51.2%, barely above the 50% expected from chance in that test. The finding does not mean every deepfake fools everyone; it shows how unreliable unaided perception can be for the material studied. Verification increasingly depends on provenance, corroboration and context, not just whether something looks real.

The synthetic-media detection challenge
51.2%

Mean classification accuracy among 1,276 study participants

Chance

50%

Study average

51.2%

Source: “As Good as a Coin Toss: Human Detection of AI-Generated Content,” Communications of the ACM. Figures reflect the study’s specific test stimuli.

AI can do more than influence what we believe

In a recent discussion with a fellow IT business owner, we moved beyond AI-generated content to a more consequential question: What happens when AI can act? The distinction is fundamental. A chatbot that summarizes an inaccurate report may mislead someone. An AI agent connected to financial software, cloud infrastructure, or administrative tools may be able to turn an inaccurate assumption into an actual transaction, configuration change, or deletion. These capabilities can create tremendous value, but their impact depends heavily on the permissions and controls surrounding them.

I believe AI has enormous potential to help businesses operate more effectively. I also believe its power warrants serious respect and a healthy amount of fear. We should be enthusiastic about what it can accomplish without pretending that speed and autonomy eliminate the need for oversight. A system authorized to read accounting data should not automatically have permission to modify financial records. A tool that can suggest a system change should not necessarily be allowed to execute it in production. Human approval, least-privilege access, tested backups and recoverable change processes are practical safeguards, not obstacles to innovation.

This is where the information problem and the automation problem intersect. AI can manufacture persuasive material, help spread it at scale and, in some deployments, carry out instructions based on what it has been told. The difference between getting a bad answer and allowing a bad answer to change the business may be nothing more than the permissions granted to the system.

How information overload becomes a business risk

Business owners live in the same information environment as their employees and customers, but their responses carry added consequences. Suppose a CEO sees a viral warning about a national infrastructure shutdown, complete with an apparently authoritative video. Without verification, the company could divert resources, disrupt operations, or create unnecessary alarm. At the other end of the spectrum, a legitimate notice about a vulnerable product could be ignored because employees have become desensitized to constant warnings. Both are failures of prioritization, not simply failures of attention.

Add impersonation and automation, and the stakes rise further. A realistic AI-generated voice message could pressure a finance employee to change payment details. A forged vendor notice could prompt staff to reveal credentials or alter security settings. A poorly constrained AI agent could perform an authorized-looking action that is inconsistent with the organization’s real intentions. These are illustrative scenarios, not claims that each happened this week, but they show why information quality belongs in the business risk conversation.

For leaders, the central question is no longer how to react to every headline. It is how to build a process that distinguishes an urgent, relevant threat from an alarming but irrelevant or unsupported story. Good risk management considers evidence, exposure, likelihood, impact, and the consequences of acting—or failing to act. It should not depend on which post generated the strongest emotional reaction.

Before a headline becomes a business decision

1. Is it verified?

Trace the claim to its original source and seek credible, independent corroboration.

2. Does it affect us?

Confirm whether your company uses the relevant vendor, service, system, or infrastructure.

3. What is the actual exposure?

Evaluate likelihood and business impact, not the urgency of the headline.

4. What response is proportionate?

Choose whether to monitor, investigate, mitigate, communicate, or take no immediate action.

5. Who is authorized to act?

Require independent verification and appropriate human approval for consequential financial, security and AI-driven actions.

Businesses do not need to turn their owners into full-time fact-checkers or security researchers. They do need trusted information sources, clearly defined decision rights, and technology partners who can translate an incident into a specific assessment of exposure and action. A threat can be real without being relevant to every organization. A story can be popular without being verified. And a technology can be powerful without being ready for unrestricted access.

Credibility is a discipline, not a feeling

My journalism background makes the erosion of confidence in information particularly troubling. Questioning what we are told is healthy; assuming that nothing can be trusted is not. Skepticism asks for evidence. Cynicism abandons the possibility of evidence altogether. When every official explanation is presumed suspicious, and every startling video can feel plausible, it becomes harder to distinguish serious reporting from manufactured certainty.

The responsibility is shared. News organizations must earn trust through accuracy and transparency; technology platforms should take the quality and context of information seriously; and businesses must build decision-making practices that can withstand urgency, uncertainty, and deception. AI raises the stakes, but it does not remove our obligation to think critically. If anything, it makes sound judgment more important.

We cannot control the volume of information competing for our attention, and we cannot prevent every strange claim from showing up in a social feed. We can control how much authority we give those claims and what we allow them to change. Business leaders cannot afford to treat every headline as an emergency, but they also cannot become so numb to the noise that they miss the threats that matter.

As for the flying jellyfish, I am still waiting for something more convincing than a viral post. At least now I am remembering to ask questions before accepting that airborne marine life has joined the list of things I need to worry about.


Sources and further reading

  1. Pew Research Center: Americans’ Complicated Relationship With News (February 2026)
  2. Pew Research Center: Digital News Platform Use (October 2026)
  3. Reuters Institute: Digital News Report 2026
  4. Reuters Institute: United States country report (2026)
  5. Communications of the ACM: As Good as a Coin Toss: Human Detection of AI-Generated Content
  6. Report URI: World’s Largest Brands Infected With Malware Via Optimizely (Nov. 16, 2026)
  7. BleepingComputer: Optimizely supply-chain attack (Nov. 16, 2026)
  8. FAA: Nationwide Ground Stop — Questions and Answers (Nov. 15, 2026)
  9. Associated Press: United Airlines outage delays departing flights (Nov. 16-17, 2026)
  10. FCC: Nationwide Emergency Alert System test notice
  11. PolitiFact: No evidence of a Nov. 17 nationwide blackout (Nov. 16, 2026)