AI-Driven Attacks Leave Defenders Less Time, IBM’s Wallin Says

Executive Summary

AI-driven attacks made up one in four malicious breaches in IBM’s 2026 Cost of a Data Breach Report and cost an average of $6 million each, IBM said on 29 July 2026. In an interview published by SecurityWeek on 13 October 2026, Steve Wallin, director of core integration and R&D labs leader at IBM UK, said AI lets attackers do familiar things much faster and at greater scale. That leaves defenders less time to respond. In Sweden, the government has asked Försvarets radioanstalt (FRA) for an interim report on an AI-supported national cyber security capability by 30 October 2026.

AI-Driven Attacks Shorten the Window for Defenders

“It just allows attackers to do the same familiar things much faster and at greater scale than ever before. The pressure comes from minimizing the time for defenders to respond,” Wallin said.

His answer starts with an organisation knowing its own security posture. “It is security by design from the ground up rather than a bolt-on reaction,” he said, naming zero trust across infrastructure, applications, API entry points and data management. Defensive AI can help with prioritisation and automation, he added. Resilience must also cover recovering damaged data through a cyber vault built on air-gapping and WORM storage; IBM offers one on its Z mainframe.

John Kindervag, who introduced zero trust in 2010, made a similar case in SecurityWeek on 1 October 2026 as his book Cyber Resilience at Machine Speed came out. An AI-generated packet still crosses the same network, he said. Correctly implemented zero trust “can still halt it”.

What IBM’s 2026 Figures Say

Three statistics in the interview read differently next to IBM’s own releases. SecurityWeek puts AI-generated attacks at little more than 20% of all attacks; IBM’s global release gives one in four malicious breaches, while IBM UK said on 28 July 2026 that 22% of British organisations reported AI-generated attacks. The 17% that SecurityWeek links to phishing matches IBM UK’s share of AI-enabled attacks that were phishing campaigns, behind deepfake impersonation at 45% and AI-enabled malware at 19%. The phishing cost appears in the interview as $5.29 per breach. eSecurity Planet, reporting on the same study, gives $5.29 million for voice and SMS phishing, the highest of any vector.

The Ponemon Institute studied 602 organisations breached between March 2025 and February 2026. The global average breach cost rose 12% to a record $4.99 million and AI-driven attacks increased 56%. Organisations with extensive security AI and automation saved $1.93 million per breach.

We found no Nordic breakdown of the 2026 figures as of 14 October 2026.

Sweden Tasks FRA With an AI Defence Pilot

A report from the Royal Swedish Academy of Engineering Sciences (IVA), written in dialogue with the National Cyber Security Centre (NCSC) and AI Sweden, said on 30 June 2026 that AI can find and exploit vulnerabilities on a scale far beyond human capability. Its priorities include faster decisions, MFA, access control, patching, logging and using AI in defence. “The most critical thing in the near term is to quickly close the most important security gaps and to integrate AI responsibly into our defences against AI-driven threats,” said Pontus Johnson, KTH professor and chair of the working group.

On 3 September 2026 the government told FRA (decision Fö2026/01313) to examine an AI-supported national capability at NCSC and start a pilot that supports critical operators in the private and public sectors. The decision names AI-supported vulnerability analysis and testing as one possible scope. The final report is due on 29 January 2027.

Vulnerability Work Is the Gap Both Sources Name

IBM X-Force’s Limor Kessem wrote on 29 July 2026 that 50% of breached organisations use AI agents in threat hunting, response and containment. Only 18% use them for vulnerability scanning and management. That is the area the Swedish decision names as a possible first pilot for FRA. FRA’s interim report on 30 October 2026 must set out the pilot’s direction and limits.

References

1. IBM Study: One in Four Malicious Breaches are AI-Enabled, Costing Companies $6 Million on Average
2. IBM Study: More Than One in Five Malicious UK Breaches are AI Generated
3. AI-powered adversaries and the enterprise risk challenge
4. Uppdrag till Försvarets radioanstalt och Försvarsmakten att analysera och påbörja utvecklingen av en AI-stödd nationell cybersäkerhetsförmåga
5. Report: AI-driven Cyber Threats – The Next Twelve Months
6. AI Turns Familiar Cyberattacks Into a Race Against Time
7. Zero Trust Creator Says Model Holds Firm Against AI-Assisted Attacks
8. IBM 2026 Cost of a Data Breach Report: Key Findings

This post is also available in: