Janvier 5, 2024

Data in Hot Water: The Cybersecurity Risks of Generative AI


Generative Artificial Intelligence (AI) is a powerful technology that can create realistic and novel content, such as images, text, audio and video. Its use cases span the enterprise, and include enhancing creativity, improving productivity, and generally making everyone’s jobs a little bit easier.


However, generative AI also poses significant cybersecurity risks to your data. From seemingly innocuous prompts that may contain sensitive information, to building large-scale malware campaigns, generative AI is near single-handedly expanding the attack surface of the modern enterprise.


These applications are here to stay, which means businesses need to adapt their security strategy to accommodate them. Here’s how to get started.

DLP for ChatGPT and Generative AI

The Three Main Security Risks with Generative AI

Generative AI security risks are multi-faceted threats that stem from how users inside and out of the organizations interact with the tools.

The three major security risks are in many respects the core tenets of cybersecurity, but their evolution in the era of AI may surprise you:

  • Data breaches: Generative AI systems can collect, store, and process large amounts of data from various sources – including user prompts. If these data sets contain sensitive information, such as unreleased financial statements or intellectual property, then enterprises open themselves up to third-party risk akin to storing data on a file-sharing platform. Tools like ChatGPT or Bard could also leak that proprietary data while answering prompts of users outside the organization.
  • Malware attacks: Generative AI can also generate new and complex types of malware that can evade conventional detection methods. Organizations may face of a wave of new zero-day attacks because of this. Without purpose-built defense mechanisms in place to prevent them from being successful, IT teams will have a difficult time keeping pace with threat actors.
  • Phishing attacks: Generative AI excels at creating convincing fake content that mimics real content but contains false or misleading information. This fake content can be used to trick users into revealing sensitive information or performing actions that compromise the security of the business. Threat actors can create new phishing campaigns – complete with believable stories, pictures and video – in minutes, and businesses will likely see a higher volume of phishing attempts because of this.


How to prevent generative AI security risks

The three main security risks stemming from generative AI all follow one common throughline: data.

Whether it’s accidental sharing of sensitive information or targeted efforts to steal it, AI is further amplifying the need for robust data security controls.

Mitigating the security risks of generative AI centers around three key concepts: employee awareness, security frameworks and technological solutions.

  • Employee awareness: Educating employees on the safe handling of sensitive information isn’t a new practice. But the introduction of new AI tools to the workforce commands attention be paid to the new data security threats that come along with it. Ensure that employees understand which information can and can’t be shared with AI-powered solutions. Similarly, make people aware of the increase in malware and phishing campaigns that may result from generative AI.
  • Security frameworks: Over the past few years, enterprises have doubled down on the Zero Trust framework. It remains an effective way of keeping data out of the hands of those who shouldn’t have it by limiting access to critical applications and preventing malware attacks before they have a chance to strike. Pairing these aspects with the benefits of Secure Access Service Edge (SASE) can give organizations greater visibility of AI application usage across the network and a stronger grasp of their data across managed and unmanaged devices.
  • Technological solutions: Technologies like Data Loss Prevention (DLP) have existed for years, enabling organizations to stop sensitive information from being copying and pasted into chatbots. Risk-Adaptive Protection (RAP) enables organizations to automate policy enforcement based on user behavior, ensuring compromised accounts are stopped from exfiltrating data. These existing tools only need fine tuning to meet emerging security demands. However, it’s likely enterprises will increasingly move toward adopting a suite of data security solutions to keep up with the ever-changing face of data breaches and leaks.


Enforce Data Security Everywhere

It’s tough to imagine a world where data isn’t the most safely guarded asset in the business. As everything around sensitive information evolves, so too must the policies and protections that are in place to keep it safe. We call it data security everywhere and we’ll keep building on the momentum we started last year.

Learn how Forcepoint is helping enterprises enforce data security policies everywhere data goes – including generative AI.


Forcepoint-authored blog posts are based on discussions with customers and additional research by our content teams.

Read more articles by Forcepoint

À propos de Forcepoint

Forcepoint est l’entreprise leader en cybersécurité pour la protection des utilisateurs et des données. Notre objectif est de protéger les entreprises tout en stimulant la transformation et la croissance numériques. Nos solutions s’adaptent en temps réel à la façon dont les personnes interagissent avec les données, et offrent un accès sécurisé tout en permettant aux employés de créer de la valeur.