Here’s what I think. My top three.
Artificial intelligence
In the world of cyber defense, AI in defense is clumsily integrated (today), fraught with false positives, but today is just the beginning. In both attack and defense, integration and usefulness of AI will get SO much better in 2025; transformative, a disrupter. There will be a gap between attack and defense. There always is. Attackers have already taken the early lead in adoption: deepfakes and dead accurate social engineering. BEC scams dwarf ransomware attacks. Why? How? AI-generated social engineering is used to convince someone to send them a check. Capabilities in attack and defend will level out, likely not in 2025 but soon after.
What’s coming next? Here's some speculation. Remember Bees with Machine Guns? Bees with Machine Guns uses numerous micro EC2 instances (the bees) to load test web applications. Think about hundreds of AI-driven self-learning micro EC2 instances attacking an entire infrastructure all at the same time. Think cyber swarms using AI to guide multi-vector high volume attack – not just DDoS; high speed overwhelming attacks. Defenses are going to need to keep up. The volley of attack and defense will be carried out at speeds no human could imagine, analyze, and correlate. Long gone are the days of dumping packet captures and running them manually. 2025 will be a significant year for AI.
Next, AI-driven Information Warfare (an old term but still accurate) against the masses is coming. “I read it on the Internet, it must be right, right?” How many times have each of us said this?! Think about that! LLMs are taught by feeding data from the Internet. Could the output of an LLM be shaped by feeding it volumes of data?
Have you noticed any of the LLMs giving you answers containing slanted product information? I asked Gemini (I love Gemini!) about correlating cyber security data. It gives me Microsoft Azure as an answer. I had to tell Gemini to answer but without Azure!
I can’t wait to see how AI shapes marketing and news. I refuse to hire analysts who use only AI (and we’ve had a few). Keep thinking independently.
What about Quantum computing?
There’s been speculation about quantum computing for years. 2025 will be the year that we see risks to existing encryption methods. Interestingly enough, we’ve seen (heard) vendors hawking “quantum-resistant cryptography” based on NIST standards. [1].
Many companies (around the world) are busy developing and offering Quantum computing, offering various levels of access: IBM, Google, Microsoft, Intel, Amazon, plus IONQ and the Chinese, Origin Wukong.
Much of this is still marketeer noise. NIST says they believe quantum computers will break encryption within the next decade. Me? We’re more than inching toward it; we’re marching, and the footsteps are growing louder.
Ransomware attacks
Ransomware is by far the biggest threat to cyber today. It will continue to be a major threat, evolving with new techniques and becoming more disruptive incorporating AI and automation, making them more sophisticated and harder to detect. This is a no-brainer. Lockbit 4 is coming out in the spring (February? March?) and others are standing in line directly behind them.
Ransomware operators will take advantage of AI. It’s cheap and easy to use. Ransomware operators building AI into their operations is a no-brainer. A stop sign could have predicted that. But what about Quantum? When Quantum is as cheap to use as AI, expect it. My guess? We’re going to measure intent by monitoring bad guys hoarding encrypted data. When we see that, we’ll know they likely intend to use Quantum computing to break encryption on previously protected data, and ransome owners. I don’t expect this in 2025, but it will come.
2025 is going to be awesome. the tech is changing so fast (again). I can't wait to see how this unfolds!
[1] https://www.federalregister.gov/documents/2024/08/14/2024-17956/announcing-issuance-of-federal-information-processing-standards-fips-fips-203-module-lattice-based
