When Fake Reads Better Than Fact
A new UK center will fight hostile disinformation. The harder problem is the decision made ten minutes later.
In an exercise linked to NATO, experienced war planners had accurate reports about a fictional region. They also saw one fake news video. Reality Defender, the firm that ran the test, says they ordered military action based on the fake and ignored the accurate reports. The company says poor training did not explain the result.[1]
Treat that with care. Reality Defender sells detection tools, and we have not seen the full method. But the result fits what NATO’s science body describes: attacks on how people think, aimed at soldiers and civilians.[2]
On September 22, UK Prime Minister Andy Burnham told the UN he was tasking security chiefs to begin work on a National Centre for Information Defence. It will detect, trace, and disrupt hostile state information attacks. He singled out Russia, and said the Kremlin spends around £1.3 billion a year on manipulating information.[3] Few details have been published yet[4], but three lessons stand out.
1. Soldiers are targets, not just audiences
Euronews reports that Russia is using AI-made videos of Ukrainian soldiers to suggest weak leadership.[5] A Defence Horizon Journal paper argues that fake media about deployed forces is now a force-protection risk. The people at risk are not only commanders. They are troops, recruits, and their families.[6]
2. Criminals now work for state campaigns
Officials describe “disposable agents,” recruited on Telegram for small paid tasks that can grow into sabotage.[7,8] A UK ringleader was jailed for 17 years over Wagner-linked arson plots.[9] A House of Commons Library briefing says MI5 has described Iranian state actors using criminals as proxies too.[10] The same tools drive fraud. UK Finance says organized groups now use deepfakes and cloned voices.[11]
3. Detection is not a decision
Spotting a fake is one job. Deciding what to do with it is another. A fake audio clip of a UK politician passed 800,000 views on X in 2024. When Full Fact looked at it, it could say only that there was no evidence the clip was real[12], it could not confirm it was 100% fake, even though the politician said it was. A national center can help find and name campaigns. It cannot stand in a briefing room at 3 a.m. and slow down a decision.
Three questions to ask now
- Who checks a source before someone acts on it?
- How fast can they do it?
- What do people do when they cannot tell — what is the failsafe scenario?
Teams that answer these before a crisis have a plan. Teams that wait will make it up on the day.
How PreEmpt.Life handles disinformation
PreEmpt.Life’s Disinformation capability works in two stages. Every claim reaching the service is first put through Deep Scan Disinformation, which builds a structured case — sourcing it, tracing how it’s spreading, reading the likely intent behind it, and scoring confidence in the finding. When a case reaches a high-confidence threshold, PreEmpt.Life doesn’t act on the AI’s read alone: it routes the case through a six-lens synthetic panel — checking verification, network behavior, tactics, media authenticity, alternative explanations, and the likely effectiveness of a correction — before a named human reviewer (one of us) confirms the case and decides what happens next.
Two safeguards run alongside that panel. An Equity & Reach Gap Check makes sure any recommended response actually reaches the community the disinformation is targeting, in the right language and through the right channel, rather than defaulting to a one-size-fits-all correction. Then a growing Signature Library cross-checks each case’s tactics against previously reviewed disinformation patterns, so PreEmpt.Life gets faster and more precise at recognizing a known method each time one is confirmed. The result is a system built to counter disinformation the way it actually spreads — quickly, but never on the AI’s judgment alone.
We poll for Disinformation once a week for free, publicly, but have built 30-minute reporting on key disinformation alerts for clients.
PreEmpt.Life can stand up always-on alerting on any topic a client names — not just disinformation — running the same continuous-detection architecture (standing scan, synthetic-panel stress test, human confirmation gate) against whatever signal matters to them.
Get Alan Gates’s stories in your inbox
Join Medium for free to get updates from this writer.
The stakes are real and cross-sector: a single hacked tweet briefly wiped $136 billion off the S&P 500 in 2013;[13] a fake insulin-pricing tweet was followed by a fall of more than 4 percent in Eli Lilly’s shares the next trading day;[14] false 5G-coronavirus claims led to arson attacks on UK telecom infrastructure.[15] Clients set the topic and the threshold; PreEmpt.Life watches it continuously and only surfaces what’s been checked, not the moment it’s spotted.
Oversight still matters
The center has critics.
Big Brother Watch backs action against foreign campaigns but wants a strictly overseas remit and independent oversight.[16] The Defence Secretary says the center will not police domestic dissent.[17]. We will see about that. Both are statements, not safeguards. Watch what gets published: its mandate, its oversight, and its evidence standards.
The state will build its center. The decision will still be made in a room where someone has ten minutes and half the facts. That is where much of the harm is done, or avoided.
ALEXIS Navigator™ already includes disinformation scanning, helping organizations, including those in defense and security, detect and interpret hostile or manipulated information from a range of threat actors, so people stay in charge of the decision.
Sources
- Biometric Update, “With NATO experiment, Reality Defender exposes military’s deepfake weakness” (vendor account) — February 11, 2026
https://www.biometricupdate.com/202602/with-nato-experiment-reality-defender-exposes-militarys-deepfake-weakness - NATO Science & Technology Organization, Chief Scientist Research Report on Cognitive Warfare — Undated on page
https://www.nato.int/content/dam/nato/webready/documents/sto/chief-scientist-report-cognitive-warfare.pdf - GOV.UK (Prime Minister’s Office), “PM speech at UNGA: 22 September 2026” — Delivered September 22, 2026; published September 23, 2026
https://www.gov.uk/government/speeches/pm-speech-at-unga-22-september-2026 - The Record, “Burnham announces plan for new UK center to fight disinformation” — September 2026
https://therecord.media/uk-disinformation-russia-center - Euronews, “Deepfakes of frontline Ukrainian soldiers aim to undermine morale” — April 22, 2026
https://www.euronews.com/my-europe/2026/04/22/deepfakes-of-frontline-ukrainian-soldiers-aim-to-undermine-morale - The Defence Horizon Journal, “AI And Microtargeting Disinformation As A Security Threat To The Protection Of International Forces” — 2026
https://tdhj.org/blog/post/ai-disinformation-security/ - RFE/RL, “Exclusive: The Russian Neo-Nazi Behind A Shadowy GRU Recruitment Campaign” — 2025
https://www.rferl.org/a/russia-gru-nazi-sabotage-recruitment-telegram/33539661.html - Propastop, “Russia’s Shadow Army in Europe: How Are Disposable Agents Recruited?” — July 9, 2026
https://www.propastop.org/en/2026/07/09/russias-shadow-army-in-europe-how-are-disposable-agents-recruited/ - Al Jazeera, “Court finds Russia’s Wagner mercenary group behind UK arson attackers” — October 25, 2025
https://www.aljazeera.com/news/2025/10/25/russia-backed-arson-attack-ringleaders-handed-hefty-jail-sentences-in-uk - House of Commons Library, “Iranian state threat activities in the UK” (Research Briefing CBP-10413) — June 16, 2026
https://researchbriefings.files.parliament.uk/documents/CBP-10413/CBP-10413.pdf - IBTimes UK, “What Is a Deepfake? UK Fraud Losses Hit £1.28B as Banks Warn AI Makes Scams Harder to Spot” (reporting UK Finance Fraud Report 2026) — 2026
https://www.ibtimes.co.uk/ai-driven-fraud-costs-uk-1-28-billion-1808188 - Full Fact, “No evidence audio clip supposedly of Wes Streeting comments about Palestinian deaths is genuine” — July 2024
https://fullfact.org/election-2024/wes-streeting-audio-clip-palestine/ - Bloomberg, “Fake Post Erasing $136 Billion Shows Markets Need Humans” (subscribers only) — April 23, 2013
https://www.bloomberg.com/news/2013-04-23/fake-report-erasing-136-billion-shows-market-s-fragility.html - Forbes, “Fake Eli Lilly Twitter Account Claims Insulin Is Free, Stock Falls 4.37%” — November 12, 2022
https://www.forbes.com/sites/brucelee/2022/11/12/fake-eli-lilly-twitter-account-claims-insulin-is-free-stock-falls-43/ - CNBC, report on UK cell towers set on fire amid 5G coronavirus conspiracy theories — April 6, 2020
https://www.cnbc.com/2020/04/06/coronavirus-uk-cell-towers-set-on-fire-amid-5g-conspiracy-theories.html - Big Brother Watch, “Big Brother Watch response to Government announcing a National Centre for Information Defence” — September 2026
https://bigbrotherwatch.org.uk/press-coverage/freespeech/big-brother-watch-response-to-government-announcing-a-national-center-for-information-defence/ - GB News, “Labour insists Britons won’t be targeted for free speech despite ‘Orwellian’ fears over new ‘information unit’” — September 2026
https://www.gbnews.com/politics/free-speech-labour-britons-wont-be-targeted-orwellian-fears-new-information-unit
Outlined with the help of AI, but written by a real human. I know . . . shocking isn’t it!.
