World War 3 and Technology: What Is Realistic?

None of the credible sources I read forecasts World War 3, but several describe the most dangerous period in years. In January 2026 the Bulletin of the Atomic Scientists set its Doomsday Clock to 85 seconds before midnight, the closest it has ever been. Technology such as drones, cyber operations, AI and satellites is changing how wars are fought and how fast a crisis can escalate, yet it does not decide on its own whether a world war happens. This post separates what is documented from what is inference and what is opinion. The date of writing is 8 November 2026, and several of these situations change week to week, so treat anything marked "as of" as a snapshot.

What do credible sources say about the risk of World War 3 in 2026?

The Bulletin of the Atomic Scientists published its 2026 Doomsday Clock statement on 27 January 2026. News coverage says the clock moved four seconds closer than the 89 seconds set in 2025. The statement points to three regional conflicts involving nuclear powers that could escalate, an arms race between major powers, the approaching end of the New START treaty, the spread of AI into military systems and AI-boosted disinformation. Its examples include Russia's war in Ukraine, the India and Pakistan clashes of May 2025 and the strikes on Iranian nuclear facilities in June 2025. The clock is a judgment by a board of scientists, so it is an expert judgment, and other sources count different things. The Institute for Economics and Peace published its Global Peace Index on 9 June 2026. It reports 61 active state-based conflicts, which it describes as more than at any point since the end of the Second World War, and says 99 of the 163 countries it covers became less peaceful over the year. Military spending points the same way. The Stockholm International Peace Research Institute (SIPRI) reported on 27 April 2026 that world military expenditure reached a record 2,887 billion dollars in 2025. That is a 2.9 percent real-terms rise, the eleventh year in a row of growth, and 2.5 percent of global GDP, the highest share since 2009. A careful reading is this. These sources describe many wars, rising spending and weaker arms control. None of the ones I read reports a direct war between the United States, Russia and China, and none predicts one. "World War 3" is a label people use for a feeling of danger, and I could not find a verified probability attached to it. Experts also disagree on how to weigh the signals above.

What happened in 2026 that people are reacting to?

The largest event is the war involving Iran. According to an Associated Press report carried by PBS NewsHour, it began on 28 February 2026 with United States and Israeli strikes on Iran. A tentative two-week ceasefire was announced on 8 April, and many issues were unresolved. The report also said Iran had effectively closed the Strait of Hormuz, through which about a fifth of the world's traded oil passes. I could not confirm the situation as of this week from a source I opened, so check live news for the current state. South Asian readers will also remember May 2025. A Congressional Research Service briefing from 13 May 2025 describes the sequence. After an attack near Pahalgam on 22 April, India launched strikes on 6 and 7 May, the sides exchanged drone, missile and artillery fire until 10 May, and a ceasefire followed. India and Pakistan still disagree about responsibility, what was hit and what was shot down. Two nuclear-armed neighbours fought with modern weapons for days, then stopped. Both halves of that sentence matter.

How are drones changing war?

Cheap drones have made precision strikes affordable. The best-documented case is Ukraine. The Kyiv Independent reported in May 2026 that small explosive drones account for about 80 percent of battlefield casualties, though the article gives no source for that number, so treat it as a rough figure. The same article says a Ukrainian interceptor drone costs roughly 2,300 dollars, while a Shahed-type attack drone costs at least 35,000 dollars. That gap helps explain why interceptors are being built in large numbers. Tactics adapt quickly. A fiber-optic cable instead of a radio link cannot be jammed, but the article says the price of a 50 km spool rose from about 300 to about 2,500 dollars, and that both sides import most of their cable from China. A cheap weapon still depends on a global supply chain. The human cost is the part to hold onto. UN High Commissioner for Human Rights Volker Türk told the Human Rights Council on 3 July 2026 that short-range drone attacks on both sides of the front line had pushed civilian casualties up between December 2025 and May 2026. My inference is that cheap drones lower the price of striking, which can make leaders more willing to use force below the level of full war. They also make it harder to tell a limited strike from the start of something bigger. Analysts debate whether drones make escalation less likely or more, and the May 2025 crisis can be read either way.

How do cyber attacks and GPS jamming fit in?

Most cyber activity in the sources I read is espionage, disruption attempts and preparation. The clearest primary document is CISA advisory AA24-038A, released by US agencies on 7 February 2024. It describes a Chinese state-sponsored group called Volt Typhoon. The agencies assessed with high confidence that the group was pre-positioning inside US communications, energy, transportation and water networks, so that it could disrupt operational technology in a future crisis. Some intrusions had lasted at least five years. That is the US government's assessment, and it describes preparation. No attack had happened. It shows why infrastructure belongs in this discussion: in a crisis, the ability to switch off a port or a power grid would matter. It also shows the difficulty of attribution, since being inside a network can be read as intent or as routine espionage. The sources I read do not document a cyberattack that started a war by itself. Satellite signals are the other soft spot. Aircraft navigation and the timing behind phone networks depend on them, and they can be jammed or spoofed with fairly small equipment. The Estonian public broadcaster ERR reported on 26 May 2026 that aircraft data showed red zones around Estonia, Finland and Kaliningrad where more than 10 percent of aircraft in the previous 24 hours had detected GPS interference. It said full protection is not achievable, so aviation and infrastructure have to plan for unreliable navigation. Delays and rerouted flights are the visible effect for ordinary people.

How is AI used in targeting and military decisions?

People mean two things by "military AI". One is decision support: software that sorts imagery and data and suggests what to look at or hit. The other is autonomy: a weapon that selects and engages targets without a human deciding on each one. Mixing them up causes most of the confusion online. On decision support, the Global Peace Index says AI has shortened targeting from about a day to seconds, and that machines make combat decisions faster than humans can review them. The National reported on 11 March 2026 that US planners use the Maven Smart System and that a commercial AI model had been added to summarise field intelligence. It quotes the US Central Command chief saying the final decision is always made by a human. Experts from Chatham House and RUSI worry about automation bias, where a human nominally approves what the machine suggests, and about pressure to move faster. The same article says a strike on a school in Iran is under a Pentagon investigation and that no role for AI has been established. Reporting on this topic often mixes confirmed facts with plausible guesses, so I am wary of any source that is sure about details the military has not published. On autonomy, the International Committee of the Red Cross (ICRC) said at a UN Geneva briefing on 25 August 2026 that there was no confirmed evidence of fully autonomous weapons being used to target human beings. It noted that some autonomous systems, such as ship-based missile defence, have existed for decades. For a wider look at where AI helps and where it harms, see my post on the future of AI, its benefits and risks.

Can autonomous weapons be regulated?

Efforts are underway, with real disagreement. The UN General Assembly passed a resolution on lethal autonomous weapons in December 2024 by 166 votes to 3, with 15 abstentions, according to Human Rights Watch. UN Secretary-General Antonio Guterres and the ICRC president have called for a legally binding instrument by 2026. They renewed the call on 25 August 2026. A UN Geneva press briefing quotes it as warning that the world is "dangerously close to crossing a moral red line: the autonomous targeting of humans by machines." It names the Seventh Review Conference of the Convention on Certain Conventional Weapons, in Geneva this month, as the clearest chance to decide whether to negotiate. A UN official said states, not the UN, decide the framework and timeline, and that talks would not happen overnight. The ICRC wants prohibitions on unpredictable systems and on systems that target people directly, plus limits on others, such as where and when they can be used. There are fair reasons for caution on the other side. Definitions are hard, some defensive systems have worked for decades, and the forum that handles this topic works by consensus, which Human Rights Watch gives as the main reason it has produced no substantive outcome since 2014. I did not find a source I could open that sets out each government's position, so I will not list them. The conference is still ahead, so the outcome is unknown.

How do disinformation, chips and rare earths matter?

The Bulletin warns that AI could supercharge false information campaigns and erode the shared facts a society needs in a crisis. The idea is not new. On 16 March 2022 a fake video of Ukrainian President Volodymyr Zelensky telling soldiers to surrender circulated online. Meta removed it as manipulated media, and Zelensky answered with his own video the same day, according to TechCrunch. Ukrainian officials had warned two weeks earlier that such fakes might appear. Generation tools have improved since, so preparing people early looks sensible. Chips and minerals are the physical side. Advanced AI hardware and modern weapons both depend on a few factories and processing plants. I covered the commercial angle in my post on Nvidia, AI chips and the risks to demand. On minerals, mining-technology.com reported in April 2026 that China suspended its October 2025 expansion of rare earth export controls for one year, with the pause due to expire on 10 November 2026. Licensing rules from April 2025 for seven medium and heavy rare earths stay in force, and the article cites GlobalData saying China processes up to 90 percent of the world's rare earths. That date is two days away, and I could not find whether the pause will be extended. My inference is that dependence on a few suppliers gives their governments bargaining power in a dispute. Whether that makes war more or less likely is an open question.

Where does nuclear risk stand after New START?

According to an AFP report in Dawn, the New START treaty expired on 5 February 2026. Its limits, 1,550 deployed strategic warheads and 800 launchers for each of Russia and the United States, no longer apply. Guterres called it a "grave moment for international peace and security." Vladimir Putin had proposed keeping the limits for another year, and Donald Trump said in October 2025 that the idea sounded good, but the report found no follow-up. Coverage of SIPRI's June 2026 yearbook said no new talks were underway, and I could not confirm anything later. The SIPRI Yearbook 2026 chapter on nuclear forces estimates about 12,187 nuclear warheads worldwide in January 2026, of which 9,745 were in military stockpiles and about 4,012 were deployed with operational forces. It says China's arsenal grew from 600 to as many as 620 warheads during 2025, and that low transparency hinders assessment of all the arsenals. The Bulletin also urges Russia, China and the United States to set guidelines on AI, particularly in nuclear command and control. I could not find a public source showing AI making nuclear decisions. The stated worry is speed and the risk of misreading warnings.

What are the realistic levels of risk?

"World War 3" bundles separate risks, and the evidence is very different for each. This is my own sorting of the sources above, not an official scale.

Text

LEVEL 1: Cyber, jamming and gray-zone incidents
  Documented: CISA advisory on pre-positioning (Feb 2024);
  GPS interference near the Baltic (May 2026); Zelensky
  deepfake (Mar 2022)
  Main uncertainty: attribution, and whether activity is
  preparation, espionage or signalling

LEVEL 2: Regional wars, including between nuclear-armed states
  Documented: 61 state-based conflicts (Global Peace Index,
  Jun 2026); Ukraine; India and Pakistan (May 2025); Iran war
  (from 28 Feb 2026)
  Main uncertainty: whether ceasefires hold and whether
  outside powers are drawn in

LEVEL 3: Direct war between great powers
  Documented: record spending (SIPRI, Apr 2026); no direct war
  in the sources I read
  Main uncertainty: miscalculation during a regional crisis

LEVEL 4: Use of nuclear weapons
  Documented: about 12,187 warheads (SIPRI, Jan 2026);
  New START expired 5 Feb 2026; clock at 85 seconds
  Main uncertainty: arms control talks, accidents, and whether
  AI enters command and control

Levels 1 and 2 are happening now. Levels 3 and 4 have not happened, and the sources warn about rising danger without predicting either. The step from a regional war to a great-power war is the one analysts watch most closely, and nothing I read says it is inevitable.

What can ordinary people and developers do?

The useful actions are small and legal. This is general information, not security, legal or professional advice.

Check before you share. Find the original source of a video or claim, look at its date, and be suspicious of content built to make you angry or afraid.
Tighten your own security with updates, a password manager, two-factor authentication or passkeys, and backups. My post on privacy, AI and surveillance has a checklist.
Learn the basics of preparedness. The European Commission's Preparedness Union Strategy, announced on 26 March 2025, asked member states to ensure citizens can be self-sufficient for at least 72 hours in a crisis. The same habits help in floods and long power cuts, so check what your local disaster authority recommends.
Support credible humanitarian organisations, and check that a charity is registered before donating.

Developers have extra options. Build software that degrades gracefully when location, connectivity or a single cloud region disappears. Use least-privilege access, as in my post on role-based access control with Next.js and Prisma, store credentials safely, as in the JWT authentication guide, and keep less data, because stored data can be breached. The apps I build, such as the GoCart marketplace, are ordinary web projects, and these basics apply to them as to anything larger.

What do I make of all this?

I am a final-year computer science student who follows the news, not a defence analyst, and I am based in Pakistan, so the May 2025 clash felt close to home. Read this section as a developer thinking aloud. More about me is on my about page. The evidence supports a few modest conclusions. Technology is making war cheaper, faster and more exposed to civilians, and it is pulling infrastructure, chips and information into the contest. It does not show that a world war is inevitable. The record so far is many regional wars, some close calls and a loss of arms-control limits, with no direct clash between the biggest powers. I find the levels above more useful than the label "World War 3", because a different response fits each one. The AI backlash is partly a reaction to these same worries about speed and accountability. I am unsure how much AI changes escalation. The reporting on AI targeting is full of claims no outsider can check, and automation bias is a plausible worry that I cannot measure. I also cannot judge how reliable each government is about its own war, so I try to lean on UN bodies, SIPRI and similar institutions and to say when a figure comes from one side. I would change my view if a direct clash between nuclear-armed great powers occurred, if the Geneva talks on autonomous weapons produced binding rules or collapsed, or if credible evidence emerged that AI was wired into nuclear decisions. I would also welcome corrections from readers who know this field better than I do.

Want to talk about this?

I am a full stack developer in Rahim Yar Khan, Pakistan, a final-year Computer Science student who works in web development at Hadi Dev Lab. If you want to talk about this topic, or about building software that handles outages and security sensibly, you can message me on WhatsApp or by email. You can also look at my work. Cover photo from Unsplash.