At the RSAC Conference 2025 in San Francisco, cybersecurity experts explored current trends and challenges in the field. Discussions highlighted the continual relevance of quantum cryptography, emphasizing the need to prepare for future threats. Cisco's findings revealed alarming risks associated with fine-tuning AI models, indicating a potential increase in vulnerabilities. Additionally, the conversation underscored that many threat actors are profit-driven individuals utilizing basic communication methods, reminding attendees of the human element in cybersecurity threats despite the technological advancements.
This isn't new territory - we were having a discussion earlier and saying that quantum cryptography has been a point of discussion at RSA for many years. It's one of those situations where you have to prepare for it, even though it might be very far in the future, and you can't easily predict when a working quantum computer is going to become an actual threat.
Cisco has apparently found that when you fine tune AI models, it can actually increase the risk of the models themselves. So Cisco research found that tailoring models, say to your unique organizational style, data, can make them three times more susceptible to jail breaks and up to 22 times more likely to output harmful responses.
All of this is to say, amidst everything we're hearing about, threat actors using AI and threat actors focusing on new methodologies, the number one thing is still profit-driven groups, that are just people. They have Discords, they have Telegram channels, they message each other, they make major mistakes.
Collection
[
|
...
]