2024 Global Digital Trust Insights

GenAI for cyber defence is on the rise

Global Digital Trust Insights
  • Issue
  • 10 Minute Read

In its 26th year, PwC’s Global Digital Trust Insights is the longest-running annual survey on cybersecurity trends. It’s also the largest survey in the cybersecurity industry, reflecting the views of over 3,800 senior security, technology and business executives.

The reinvention and innovation that businesses are doing today connect more digital experiences using the latest tech tools. Cybersecurity should be right there at the epicentre, hence the theme of our 2024 survey. We have a C-suite playbook for those who dare to break cyber-as-usual.

Special feature from the survey GenAI for cyber defence is on the rise

Don’t lose sight of governance amid all the excitement

  • A bit more than half (51%) of Swiss companies say they’ll use generative AI (GenAI) for cyber defence in the next 12 months, according to the 2024 Global Digital Trust Insights survey. On a global level, seven in 10 senior executives (69%) already plan the use of GenAI.
  • Another surge in cyber threats may be coming because GenAI can help create advanced business email compromise at scale. CISOs and CIOs should pay attention to a prevailing sentiment: 46% of Swiss respondents expect GenAI to lead to catastrophic cyber attacks in the next 12 months.
  • Companies need to establish sound AI governance and get ahead of risks that could come from exploration with GenAI. Sixty-three percent of senior execs on a global level feel personally comfortable using GenAI tools even without data governance policies in place - in Switzerland, still 46% agree.

Generative AI is opening frontiers that more than 3,800 C-level business and tech executives who responded to our 2024 Global Digital Trust Insights (DTI) Survey are exploring — in the business and for cyber defence.

Nearly seven in 10 say their organisation will use GenAI for cyber defence. Platforms are licensing their large language models (LLMs) in tandem with their cyber tech solutions. Microsoft Security Copilot intends to provide GenAI features for security posture management, incident response and security reporting. Google announced Security AI Workbench for similar use cases and many other security vendors such as Crowdstrike and Zscaler have announced features using GenAI. Even without vendor tools, some companies have been using GenAI to identify and triage phishing attempts.


GenAI for cyber defence in Switzerland

51%

 

A bit more than half say they’ll use GenAI for cyber defence in the next 12 months.

 

25%

 

One fourth are using GenAI for cyber risk detection and mitigation.

 

11%

 

One-tenth are already seeing benefits to their cyber programmes because of GenAI — mere months after its public debut.

 

Question: To what extent do you agree or disagree with the following statements about Generative AI?
Base: 3,876 survey respondents | 72 Swiss respondents
Source: PwC, 2024 Global Digital Trust Insights.


 

GenAI comes at an opportune time in cybersecurity.

For defence. Organisations have long been overwhelmed by the sheer number and complexity of human-led cyberattacks, both of which continually increase. And GenAI is making it easier to conduct complex cyber attacks at scale. Researchers found a 135% increase in novel social engineering attacks in just one month, from January to February 2023. Services like WormGPT and FraudGPT are enabling credential phishing and highly personalised business email compromise.

To secure innovation. Businesses eager to reap GenAI’s many potential benefits to develop new lines of business and increase employee productivity invite serious risks to privacy, cybersecurity, regulatory compliance, third-party relationships, legal obligations and intellectual property. So to get the most benefit from this groundbreaking technology, organisations should manage the wide array of risks it poses in a way that considers the business as a whole.

Question: To what extent do you agree or disagree with the following statements about Generative AI?
Base: 72 Swiss respondents
Source: PwC, 2024 Global Digital Trust Insights.

The promise of GenAI for cyber defence

From reconnaissance to action, GenAI can be useful for defence all along the cyber kill chain. Here are the three most promising areas.

Threat detection and analysis. GenAI can be invaluable for proactively detecting vulnerability exploits, rapidly assessing their extent — what’s at risk, what’s already compromised and what the damages are — and presenting tried-and-true options for defence and remediation. GenAI can identify patterns, anomalies and indicators of compromise that elude traditional signature-based detection systems.

GenAI is strong at synthesising voluminous data on a cyber incident from multiple systems and sources to help teams understand what has happened. It can present complex threats in easy-to-understand language, advise on mitigation strategies and help with searches and investigations.

Cyber risk and incident reporting. GenAI also promises to make cyber risk and incident reporting much simpler. Vendors already are working on this capability. With the help of natural language processing (NLP), GenAI can turn technical data into concise content that non-technical people can understand. It can help with incident response reporting, threat intelligence, risk assessments, audits and regulatory compliance. And it can present its recommendations in terms that anyone can understand, even translating confounding graphs into simple text. GenAI could also be trained to create templates for comparisons to industry standards and leading practices.

GenAI’s reporting capabilities should prove invaluable in this new era of heightened cyber transparency. To wit: A recent law will soon require critical infrastructure entities in the US to report cyber incidents. Also, the Securities and Exchange Commission (SEC) has released rules requiring disclosures of material cyber incidents and material cyber risks in SEC filings. The European Union’s Digital Operational Resilience Act calls for timely and consistent reporting of incidents that affect financial entities’ information and communication technologies. Imagine having a tool that makes preparing these reports much easier.

Adaptive controls. Securing the cloud and software supply chain requires constant updates in security policies and controls — a daunting task today. Machine learning algorithms and GenAI tools could soon recommend, assess and draft security policies that are tailored to an organisation's threat profile, technologies and business objectives. These tools could test and confirm that policies are holistic throughout the IT environment. Within a zero trust environment, GenAI can automate and continually assess and assign risk scores for endpoints, and review access requests and permissions. An adaptive approach, powered by GenAI tools, can help organisations better respond to evolving threats and stay secure.

And more. Many vendors are pushing the limits of GenAI, testing what’s possible. As the technology improves and matures, we’ll see many more uses for it in cyber defence. It could be some time, however, before we see “defenceGPT’s” broad-scale use.

Invest in your security teams

GenAI tools could help relieve the acute cyber talent shortage. Attrition is a growing problem for 39% of CISOs, CIOs and CTOs, according to our 2023 Global DTI survey. It’s hindering progress on cyber goals for another 15%.

Once GenAI frees security professionals from routine and mundane tasks such as detection and analysis, they may turn their focus to understanding — not just knowing — the causes of breaches and how best to respond to them. They can be better positioned to make fast decisions and take swift actions. ​​They might cultivate true “deep learning” — in a human sense — of LLMs, and use them to invent new ways to secure the enterprise.

And they’ll be well equipped to pivot from finding answers — GenAI’s purview, now — to asking more meaningful questions not only of their AI models but also of one another, sparking imagination and insights that are truly new. You can help your security teams develop traits that AI won’t learn or automate, such as curiosity, empathy and intuition.

Brace for regulatory uncertainty

The use of GenAI for cyber defence — just like the use of GenAI across the business — will be affected by AI regulations, particularly concerning bias, discrimination, misinformation and unethical uses. Recent directives including the Blueprint for AI Bill of Rights from the White House and the draft European Union AI Act emphasise ethical AI. Policymakers around the world are scrambling to set limits and increase accountability — treating generative AI with urgency because of its potential for affecting broad swathes of society profoundly and rapidly.

Savvy enterprises will want to get ahead of AI mandates. Our respondents are well aware of their imminence: they’ve told us that AI regulations, more than any other, could significantly affect their future revenue growth.

Among the 37% of respondents anticipating AI regulation, three-quarters think the costs of compliance will also be significant. About two-fifths say they’ll need to make major changes in the business to comply.

 

 

Amid regulatory uncertainty, companies can control one thing: how they deploy GenAI in a responsible way in their environments, which can position themselves for compliance. Seven major developers of LLMs are showing the way. At the heart of a voluntary pledge they recently signed with the US government is an agreement to start placing guardrails around the technology’s capabilities.

Channel your enthusiasm into trusted, ethical practices

Enthusiasm for AI is so high that globally, 63% of our executive respondents said they’d personally feel comfortable launching GenAI tools in the workplace without any internal controls for data quality and governance. Senior execs in the business are even more so inclined than the tech and security execs. Swiss execs are a bit more conservative in their assessment, but 43% would feel comfortable still to launch tools without internal controls.

However, without governance, adoption of GenAI tools opens organisations to privacy risks and more. What if someone includes proprietary information in a GenAI prompt? And without training in how to properly evaluate outputs, people might base recommendations on invented data or biased prompts.

Employees also need to be on guard against prompt injection risks, which Open Source Foundation for Application Security (OWASP) highlighted as the top security risk related to using LLMs. Prompt injections, also called jailbreaks, refer to prompts designed to elicit unintended responses by LLMs by overwriting system prompts or manipulating inputs from external sources.

Question: To what extent do you agree or disagree with the following statements about Generative AI?
Base: 72 Swiss respondents
Source: PwC, 2024 Global Digital Trust Insights.

 

The place to start with GenAI — as with almost any technology — is by laying the foundation for trust in its design, its function and its outputs. This foundation begins with governance, but concentrating on data governance and security concerns is especially important. The lion’s share of respondents overall say they intend to use GenAI in an ethical and responsible way: 77% agree with this statement.

 

Question: To what extent do you agree or disagree with the following statements about Generative AI?
Base: 72 Swiss respondents
Source: PwC, 2024 Global Digital Trust Insights.


Don’t overlook people

GenAI tools will be able to quickly synthesise information from multiple sources to aid in human decision-making. And, given that 74% of breaches reportedly involve humans, governance of AI for defence ought to include a human element as well.

Enterprises would do well to adopt a responsible AI toolkit, such as PwC’s, to guide the organisation’s trusted, ethical use of AI. Although it’s often considered a function of technology, human supervision and intervention are also essential to AI’s highest and ideal uses.

Ultimately, the promise of generative AI rests with people. Every savvy user can — should — be a steward of trust. Invest in them to know the risks of using the technology as assistant, co-pilot or tutor. Encourage them to critically evaluate the outputs of generative AI models in line with your enterprise risk guardrails. Rally security professionals to follow responsible AI principles.

See how you stack up against your peers

Leverage the cybersecurity and privacy benchmarking tool to gain real-time insights on how your organisation is performing.

Register to download the report

Sign up to get the full report with more of the latest findings and what lies ahead for 2024.

The 2024 Global Digital Trust Insights is a survey of 3,876 business, technology, and security executives (CEOs, corporate directors, CFOs, CISOs, CIOs, and C-Suite officers) conducted in the May through July 2023 period.

Four out of 10 executives are in large companies with $5 billion or more in revenues. Importantly, 30% are in companies with $10 billion or more in revenues.

Respondents operate in a range of industries, including industrial manufacturing (20%), financial services (20%), tech, media, telecom (19%), retail and consumer markets (17%), energy, utilities, and resources (11%), health (9%) and government and public services (3%).

Respondents are based in 71 countries. The regional breakdown is Western Europe (32%), North America (28%), Asia Pacific (18%), Latin America (10%), Eastern Europe (5%), Africa (4%) and Middle East (3%). 72 respondents are based in Switzerland.

The Global Digital Trust Insights Survey had been known as the Global State of Information Security Survey (GSISS). In its 26th year, it’s the longest-running annual survey on cybersecurity trends. It’s also the largest survey in the cybersecurity industry and the only one that draws participation from senior business executives, not just security and technology executives.

PwC Research, PwC’s global Centre of Excellence for market research and insight, conducted this survey.

A Swiss perspective, authored by:

Urs Küderli

Partner and Leader Cybersecurity and Privacy, PwC Switzerland

+41 58 792 42 21

Email

Johannes Dohren

Partner, Cybersecurity and Privacy, PwC Switzerland

+41 58 792 22 20

Email

Yan Borboën

Partner, Leader Digital Assurance and Cybersecurity & Privacy, PwC Switzerland

+41 58 792 84 59

Email