Cybersecurity Trends for 2025 and Beyond

Summary

In this video, the presenter reviews cybersecurity predictions made in previous years and evaluates their accuracy. Predictions about the adoption of passkeys, AI-driven phishing, deepfake fraud, and generative AI's role in cybersecurity are explored, with most predictions proving accurate. Looking ahead to 2025, the presenter anticipates continued challenges, such as the rise of shadow AI, deeper risks from deepfakes, generative AI's role in creating malware, and the need for quantum-safe cryptography. The video emphasizes both the threats and the potential of AI in shaping the future of cybersecurity.

Takeaways

• 😀 The adoption of passkeys over passwords saw significant growth in 2024, with more users and companies moving to more secure alternatives like FIDO passkeys.
• 😀 Generative AI has become a key player in creating more sophisticated and personalized phishing attacks, making them harder to detect.
• 😀 Deepfake technology continues to improve, with notable incidents involving impersonations, such as a $25 million fraud involving a fake CFO and a deepfake robocall in an election.
• 😀 AI hallucinations are still a problem, as generative AI sometimes provides inaccurate information or makes mistakes in calculations, despite improvements.
• 😀 Securing AI deployments has become a top priority for organizations, as AI systems are vulnerable to attacks that could compromise their integrity.
• 😀 Shadow AI, or unauthorized AI deployments, will increase as employees use AI tools without approval, potentially leading to security risks like data leakage.
• 😀 Deepfake technology will continue to pose risks to businesses and governments, potentially influencing public opinion or causing legal complications by creating false evidence.
• 😀 Generative AI's ability to write malware is a growing concern, as cybercriminals can now generate exploit code with ease, increasing the frequency of cyberattacks.
• 😀 The expanding attack surface of AI systems means that attackers will target the AI itself, looking for vulnerabilities that can be exploited to cause damage or steal data.
• 😀 Prompt injection attacks will become a primary concern as malicious actors try to manipulate AI models, bypassing their safeguards to exploit them for harmful purposes.
• 😀 AI has potential to enhance cybersecurity by providing expert advice and helping cybersecurity teams respond more effectively to threats, although caution is needed to avoid relying solely on AI for automated responses.
• 😀 Quantum computing presents both an exciting opportunity and a significant risk, as it could one day break current cryptographic methods, making quantum-safe cryptography a critical need for the future.

Q & A

What was the prediction made in the 2024 video about passkeys, and did it come true?

The prediction was that passkeys would replace passwords as a more sophisticated and secure alternative. The prediction came true as the adoption of passkeys increased, with 1 in 3 users storing passkeys and twice as many websites accepting them.

How has generative AI influenced phishing attacks, according to the script?

Generative AI has made phishing emails more personalized and convincing by generating highly targeted content that looks legitimate. This has made it more difficult to detect phishing emails, as they no longer contain common errors like spelling mistakes.

What was the impact of deepfakes in cybersecurity, as discussed in the video?

Deepfakes have been used in real-world attacks, such as an incident where a deepfake of a CFO was used to trick an employee into transferring $25 million. This shows how deepfakes can be used to impersonate individuals in both business and political settings.

What is the current issue with hallucinations in generative AI, and how does it affect its accuracy?

Hallucinations in generative AI refer to instances where the AI produces inaccurate or fabricated information. An example provided was a chatbot incorrectly converting a running pace from kilometers to miles. While improvements are being made, hallucinations still pose a significant problem.

Why is securing AI deployments considered a critical issue, according to the video?

Securing AI deployments is critical because as companies use AI in various applications, ensuring these deployments are robust against attacks is a growing concern. The script highlighted that this is currently the top question companies ask cybersecurity experts.

What is 'shadow AI,' and why is it a concern for cybersecurity?

Shadow AI refers to unapproved or unauthorized AI deployments within an organization. It can lead to data leakage, security vulnerabilities, and potential misinformation, as these AI systems may not be properly secured or monitored.

How will deepfake technology continue to affect business and government?

Deepfake technology will continue to pose a threat to businesses and governments by allowing bad actors to impersonate key figures. This could lead to financial fraud, misinformation campaigns, and challenges in verifying the authenticity of video and audio evidence.

What role does generative AI play in writing malware, and how has it already been used in attacks?

Generative AI has been shown to be capable of writing exploit code, making it easier for attackers to launch malware attacks without knowing how to write code themselves. This has already contributed to a significant increase in cyberattacks, as reported by a major online retailer.

What are the security risks associated with the expanding attack surface due to AI technologies?

The expansion of the attack surface due to AI technologies means that the AI systems themselves become targets. Attackers could exploit vulnerabilities in AI models or use shadow AI to access sensitive data or disrupt business operations.

What is prompt injection, and why is it a major concern for large language models?

Prompt injection is a technique where attackers manipulate the input to generative AI systems to make them perform unintended actions. This can break through the AI's guardrails, allowing malicious actors to bypass security and control the AI's output, posing significant risks to cybersecurity.

How could quantum computers affect current cryptographic systems, and what is being done to address this threat?

Quantum computers have the potential to break current cryptographic systems, posing a major risk to data security. To prepare for this, organizations are moving toward quantum-safe cryptography, which is resistant to quantum computing attacks. However, many organizations have yet to begin this transition.

What potential positive uses of generative AI in cybersecurity were mentioned in the video?

Generative AI can assist in cybersecurity by providing expert advice on incident response and generating suggestions for mitigating cyber threats. However, automation of responses is cautioned against due to concerns about hallucinations affecting accuracy.