Cybersecurity predictions: what was said, and what happened

    91 forecasts from respected sources such as Gartner, NIST, Microsoft, the UK NCSC and Cybersecurity Ventures, newest first, with 2027 at the top. Each is marked correct, mixed, wrong or pending. My earliest sourced entry is 1991, and I could not source anything for 1987–1990.

    13correct
    17mixed
    12wrong
    49pending
    31%correct of resolved
    21%Threats & attacks3 correct · 5 mixed · 6 wrong · 3 pending
    22%Defense & tools2 correct · 4 mixed · 3 wrong · 11 pending
    50%AI & emerging tech2 correct · 2 mixed · 0 wrong · 15 pending
    38%Policy & workforce3 correct · 2 mixed · 3 wrong · 19 pending
    43%Cyberwar3 correct · 4 mixed · 0 wrong · 1 pending

    Notable correct calls

    • CERT-In / SISA Digital Threat Report (2024) Threats & attacks
    • FBI IC3 (public service announcement) (2024) AI & emerging tech
    • Cisco (2017) Threats & attacks
    • Gartner (2017) Policy & workforce
    • NIST (2016) Defense & tools
    • Cybersecurity Ventures (2016) Policy & workforce
    • Mikko Hyppönen (F-Secure) (2014) Threats & attacks
    • Dmitri Alperovitch (McAfee) (2011) Cyberwar
    • Stuxnet analysts (Symantec, Langner) (2010) Cyberwar
    • Idaho National Lab (Aurora test) (2007) Cyberwar
    • Staniford, Paxson & Weaver (2003) AI & emerging tech
    • Bill Gates (2002) Policy & workforce
    • Bruce Schneier (2000) Defense & tools

    2027

    Made as of 21 November 2026. Outcomes are not known yet.

    Pending2027

    Gartner

    AI & emerging tech

    17% of cyberattacks and data leaks will involve generative AI.

    Based on a 2024 prediction. Outcome known after 2027.

    Pending2027

    Gartner

    AI & emerging tech

    AI agents will halve the time needed to exploit account exposures.

    Predicted in March 2025.

    Pending2027

    Gartner

    AI & emerging tech

    Over 40% of AI-related data breaches will come from cross-border GenAI misuse.

    Predicted in February 2025.

    Pending2027

    Gartner

    AI & emerging tech

    More than 70% of all digital content will be synthetic.

    Presented at Gartner's September 2026 summit.

    Pending2027

    Gartner

    AI & emerging tech

    50% of enterprises will invest in disinformation security, up from under 5%.

    Predicted in 2025.

    Pending2027

    Gartner

    AI & emerging tech

    The market for securing AI will reach about $4.8 billion, up 68.7%.

    Forecast of August 2026.

    Pending2027

    Gartner

    Policy & workforce

    30% of organisations will require sovereign cloud security controls.

    Linked to geopolitical turmoil.

    Pending2027

    Gartner

    Policy & workforce

    Two-thirds of the global 100 will extend D&O insurance to cybersecurity leaders.

    Predicted in 2024.

    Pending2027

    Gartner

    AI & emerging tech

    50% of CISOs will adopt human-centric security design.

    Predicted in 2023.

    Pending2027

    Gartner

    Defense & tools

    70% of organisations will combine DLP and insider-risk tools with IAM context.

    Predicted in 2024.

    Pending2027

    Gartner

    Defense & tools

    30% of security functions will redesign app security for non-experts.

    Predicted in 2024.

    Pending2027

    Gartner

    Threats & attacks

    75% of employees will create or change technology outside IT's view.

    Predicted in 2023.

    Pending2027

    Gartner

    Policy & workforce

    AI governance will be required by all sovereign AI laws worldwide.

    Predicted in February 2025.

    Pending2027

    Gartner

    Policy & workforce

    Manual AI compliance will expose 75% of regulated firms to fines above 5% of revenue.

    Predicted in March 2026.

    2026

    Pending2026

    Gartner

    AI & emerging tech

    AI applications will drive 50% of incident-response effort by 2028.

    Predicted in March 2026.

    Pending2026

    Gartner

    AI & emerging tech

    Over half of successful attacks on AI agents will exploit access control or prompt injection by 2029.

    Predicted in August 2026.

    Pending2026

    Gartner

    Defense & tools

    GenAI with platform-based programmes will cut employee-driven incidents by 40%.

    Predicted in 2025. Outcome not confirmed.

    Pending2026

    Gartner

    AI & emerging tech

    AI TRiSM controls will cut inaccurate or illegitimate information by 50%.

    Predicted in 2025. Outcome not confirmed.

    2025

    Pending2025

    Gartner

    Threats & attacks

    40% of social engineering attacks will target executives and the workforce by 2028.

    Predicted in March 2025.

    Pending2025

    Gartner

    Defense & tools

    Preemptive cybersecurity will be 50% of IT security spending by 2030.

    From memory. Verify the exact wording.

    2024

    Pending2024

    Gartner

    AI & emerging tech

    25% of enterprise breaches will trace to AI agent abuse by 2028.

    From memory. Verify the exact wording.

    Pending2024

    UK NCSC

    AI & emerging tech

    AI will increase the volume and impact of cyberattacks over two years.

    Assessment of January 2024. The window ends in 2026.

    Mixed2024

    Many security researchers

    AI & emerging tech

    AI deepfakes would disrupt the 2024 elections.

    Misuse occurred, but measurable effects were limited.

    Pending2024

    Gartner

    Policy & workforce

    Global security spending would grow 15% in 2025.

    Forecast of August 2024. I did not confirm the actual figure.

    Pending2024

    Gartner

    AI & emerging tech

    30% of enterprises will find identity checks unreliable alone because of deepfakes by 2026.

    From memory. Verify the exact wording.

    Correct2024

    CERT-In / SISA Digital Threat Report

    Threats & attacks

    Seven threats to India's financial sector in the coming year.

    The 2025–26 report said six of the seven came true.

    Pending2024

    Gartner (via SecurityBrief)

    AI & emerging tech

    GenAI will remove the need for specialist education from 50% of entry-level security jobs by 2028.

    Outcome known after 2028.

    Pending2024

    Gartner (via SecurityBrief)

    Policy & workforce

    Enterprises will spend over $500 billion combating malinformation by 2028.

    Outcome known after 2028.

    Pending2024

    Gartner (via MSSP Alert)

    Defense & tools

    Through 2026, 75% of organisations will leave legacy and cyber-physical systems out of zero trust.

    The window ends in 2026.

    Correct2024

    FBI IC3 (public service announcement)

    AI & emerging tech

    Criminals would use generative AI to scale up fraud.

    The FBI's 2025 report logged over 22,000 AI-related complaints and $893 million in losses.

    2023

    Pending2023

    US Bureau of Labor Statistics

    Policy & workforce

    Information security analyst jobs would grow about 32% from 2022 to 2032.

    Window runs to 2032.

    Mixed2023

    Many security vendors

    AI & emerging tech

    ChatGPT would let novices write novel malware.

    Phishing and scripting help grew, but little truly novel malware.

    Pending2023

    Gartner

    Policy & workforce

    Talent gaps and human error would cause over half of major incidents by 2025.

    Outcome not confirmed.

    Pending2023

    Gartner (press release, Mar 2023)

    Policy & workforce

    70% of boards will include a member with cybersecurity expertise by 2026.

    The year is not finished.

    Pending2023

    Gartner (press release, Mar 2023)

    Policy & workforce

    50% of cyber leaders will have tried and failed to use risk quantification for decisions by 2025.

    Outcome not confirmed.

    Pending2023

    Gartner (via TechRepublic)

    Policy & workforce

    Nearly half of cyber leaders will change jobs by 2025, and a quarter will leave the field.

    Outcome not confirmed.

    Pending2023

    Gartner (Sep 2023 summit)

    Defense & tools

    Through 2026, over half of cyberattacks will target areas zero trust cannot cover.

    The window ends in 2026.

    Pending2023

    Gartner (via Technology Decisions)

    Policy & workforce

    Modern privacy rules will cover most consumer data by 2024.

    Outcome not confirmed.

    2022

    Mixed2022

    Many officials and analysts

    Cyberwar

    Russia's invasion would bring devastating cyberattacks on the West.

    There were serious attacks (e.g. Viasat), but no Western collapse.

    Mixed2022

    FIDO Alliance and major platforms

    Defense & tools

    Passkeys would replace passwords.

    Adoption is growing, but passwords remain widespread.

    Pending2022

    Gartner

    Policy & workforce

    45% of organisations will face software supply-chain attacks by 2025.

    The figure has not been independently confirmed.

    Pending2022

    Gartner

    Defense & tools

    10% of large enterprises will have mature zero trust by 2026, from under 1%.

    Outcome not confirmed.

    Pending2022

    Gartner

    Policy & workforce

    40% of boards will have a cybersecurity committee by 2025.

    Outcome not confirmed.

    Pending2022

    Gartner

    Policy & workforce

    50% of C-level executives will have cyber-risk performance goals by 2026.

    Outcome not confirmed.

    Pending2022

    Gartner

    Defense & tools

    Cybersecurity mesh would cut incident costs by about 90% by 2024.

    Outcome not confirmed.

    Wrong2022

    Gartner

    Threats & attacks

    30% of nations would regulate ransomware payments by 2025.

    A few countries and states did, far fewer than 30%. From memory.

    Mixed2022

    US OMB (M-22-09)

    Policy & workforce

    Federal agencies would reach zero-trust goals by end of FY2024.

    Progress was uneven, and many agencies fell short.

    Pending2022

    NSA (CNSA 2.0)

    Policy & workforce

    National security systems would move to quantum-resistant crypto by 2035.

    Deadline is 2035.

    Pending2022

    Cyber Safety Review Board

    Threats & attacks

    Log4j would stay exploitable for a decade or more.

    Still vulnerable systems were being found in 2024.

    Pending2022

    Gartner (via Dark Reading)

    Defense & tools

    60% of organisations will adopt zero trust as a starting point by 2025, and over half will fail to get its benefits.

    Outcome not confirmed.

    Pending2022

    Gartner (via Dark Reading)

    Policy & workforce

    60% of organisations will use cyber risk as a main factor in third-party deals by 2025.

    Outcome not confirmed.

    2021

    Pending2021

    Gartner

    Cyberwar

    Attackers will weaponise OT to harm or kill people by 2025.

    I found no verified case.

    Pending2021

    Gartner

    Policy & workforce

    70% of CEOs would require a culture of organisational resilience by 2025.

    Outcome not confirmed.

    Pending2021

    Gartner

    Defense & tools

    80% of enterprises would unify web, cloud and app access under one vendor by 2025.

    Outcome not confirmed.

    2020

    Pending2020

    Gartner

    Policy & workforce

    75% of the world's population would have privacy-law coverage by 2024.

    Outcome not confirmed.

    Pending2020

    Gartner

    Defense & tools

    40% of enterprises would have SASE strategies by 2024.

    Outcome not confirmed.

    Mixed2020

    Cybersecurity Ventures

    Threats & attacks

    Cybercrime would cost the world $10.5 trillion a year by 2025.

    Unverifiable. The FBI's 2025 report counted a record $20.9 billion in US reported losses, which is not comparable.

    2017

    Correct2017

    Cisco

    Threats & attacks

    'Destruction of service' attacks would arrive, hitting backups too.

    NotPetya (June 2017) was a wiper disguised as ransomware.

    Mixed2017

    Cybersecurity Ventures

    Threats & attacks

    Ransomware costs of $11.5B in 2019 and $20B in 2021.

    Estimates vary widely and no audited total exists.

    Wrong2017

    Brad Smith (Microsoft)

    Policy & workforce

    The world needs a 'Digital Geneva Convention'.

    Not adopted. Only non-binding efforts followed.

    Correct2017

    Gartner

    Policy & workforce

    Over half of firms in GDPR scope would miss compliance by end-2018.

    Surveys found many still not fully compliant.

    2016

    Correct2016

    NIST

    Defense & tools

    Post-quantum encryption standards would be ready around 2022–24.

    The first standards were published in August 2024.

    Correct2016

    Cybersecurity Ventures

    Policy & workforce

    3.5 million unfilled cybersecurity jobs by 2021.

    The ISC2 2022 study estimated a gap of about 3.4 million.

    Mixed2016

    Cybersecurity Ventures

    Policy & workforce

    Cumulative security spending of over $1 trillion from 2017 to 2021.

    Widely cited, but figures vary by definition.

    2015

    Mixed2015

    Cybersecurity Ventures

    Threats & attacks

    Cybercrime would cost $6 trillion a year by 2021.

    Widely cited, but disputed. The FBI's 2025 report counted a record $20.9 billion in US reported losses.

    Mixed2015

    Gartner

    Defense & tools

    Through 2020, 99% of exploited vulnerabilities would be ones known for over a year.

    Many exploited flaws are old, though zero-days have grown.

    Wrong2015

    ISC2 / Frost & Sullivan

    Policy & workforce

    A shortage of 1.5 million security workers by 2020.

    ISC2's 2020 study put the gap at about 3 million.

    Mixed2015

    US–China agreement

    Cyberwar

    China would stop hacking for commercial gain.

    FireEye saw a drop in 2016, then a return to activity.

    Mixed2015

    Juniper Research

    Threats & attacks

    Breaches would cost the world $2.1 trillion by 2019.

    Unverifiable. Estimates differ widely.

    Pending2015

    Michele Mosca

    AI & emerging tech

    A 1-in-7 chance of breaking RSA-2048 by 2026, and 1-in-2 by 2031.

    No such break has been shown so far.

    2014

    Mixed2014

    Symantec

    Defense & tools

    Antivirus was 'dead' as a sole defence.

    Antivirus survived by evolving into endpoint detection and response.

    Correct2014

    Mikko Hyppönen (F-Secure)

    Threats & attacks

    Smart devices would be a major new attack surface.

    The 2016 Mirai IoT botnet proved it.

    2013

    Mixed2013

    Gartner

    Threats & attacks

    26 billion IoT devices would be installed by 2020.

    Later counts were well below that, depending on the definition.

    2012

    Mixed2012

    Leon Panetta

    Cyberwar

    A 'cyber Pearl Harbor' was coming.

    There were major incidents (NotPetya, Colonial Pipeline), but no single Pearl Harbor-scale event.

    2011

    Wrong2011

    Cisco

    Threats & attacks

    50 billion connected devices by 2020.

    Later counts were well below that.

    Correct2011

    Dmitri Alperovitch (McAfee)

    Cyberwar

    All large companies have been or will be hacked.

    Breach disclosures since then bear this out.

    2010

    Correct2010

    Stuxnet analysts (Symantec, Langner)

    Cyberwar

    Cyberweapons would be copied and spread.

    Duqu, Flame and NotPetya followed.

    2009

    Wrong2009

    Security vendors and press

    Threats & attacks

    Conficker would strike massively on April 1, 2009.

    Little happened.

    2007

    Correct2007

    Idaho National Lab (Aurora test)

    Cyberwar

    A cyberattack could physically destroy power equipment.

    Stuxnet (2010) and Ukraine's grid (2015) did exactly that.

    2004

    Wrong2004

    Bill Gates

    Threats & attacks

    Spam would be solved by 2006.

    Spam still made up around half of email.

    Wrong2004

    Bill Gates

    Defense & tools

    Passwords were on their way out.

    Passwords remain the main login method.

    2003

    Mixed2003

    Gartner

    Defense & tools

    Intrusion detection was a failure and would be replaced by prevention.

    IDS lived on, merged with intrusion prevention.

    Correct2003

    Staniford, Paxson & Weaver

    AI & emerging tech

    Worms could infect most vulnerable hosts in minutes.

    Slammer (Jan 2003) infected most in about 10 minutes.

    2002

    Correct2002

    Bill Gates

    Policy & workforce

    Trustworthy Computing: security would become Microsoft's top priority.

    Microsoft's security development lifecycle followed.

    2000

    Correct2000

    Bruce Schneier

    Defense & tools

    Security fails through people and process, not cryptography.

    Verizon's 2024 DBIR found a human element in about 68% of breaches.

    1998

    Wrong1998

    Many analysts

    Threats & attacks

    Y2K would cause widespread infrastructure failures.

    Few failures, though remediation cost hundreds of billions.

    Wrong1998

    US PDD-63 (Clinton)

    Policy & workforce

    Critical infrastructure would be protected by 2003.

    The goal was not met.

    1995

    Wrong1995

    Critics of the SATAN scanner

    Defense & tools

    A free vulnerability scanner would cause mass break-ins.

    No catastrophe, and such tools became standard.

    1993

    Wrong1993

    US government (Clipper chip)

    Defense & tools

    Clipper would become the standard for encrypted phones.

    It was abandoned by 1996.

    1992

    Wrong1992

    John McAfee

    Threats & attacks

    The Michelangelo virus would hit millions of computers.

    Only a small fraction of that were affected.

    1991

    Mixed1991

    Winn Schwartau

    Cyberwar

    An 'electronic Pearl Harbor' was coming.

    Big attacks came, but no single Pearl Harbor-scale event.

    This is a curated sample, not a survey of all forecasts, and percentages describe this list only. Verdicts are summary judgments. The 2022–2027 Gartner entries and the FBI figures come from published sources I checked. Older entries are from memory, so verify them before citing.