Interview, Fireside Chat, Conference Presentation, Webinar
How to spot an AI Deepfake
- Email protection systems are expected to evolve rapidly to outsmart threats, yet other communication channels face significant exposure, with overall attack velocity accelerating post-DeepSeek open-source release as adversaries leverage these tools.
- Social engineering incidents have risen over fourfold since ChatGPT's launch, accompanied by a surge in sophisticated deepfakes, with over 100,000 recorded in the US in 2024 and an expectation that 30-40% of the population will face such attacks within the next year.
- Open-source model accessibility enables sophisticated, low-cost attacks via smartphones from any country, expanding vectors to include voice, SMS, video, and chat, thereby making destructive capabilities and scalable autonomous agent operations cost-prohibitive barriers for attackers.
- Generative AI is projected to significantly enhance corporate scams like CEO imposters by exploiting organizational authority structures, while news regarding major attacks is expected to occur every few days, with a feared transition from financial loss to physical harm targeting hospitals and energy systems.
- Large-scale attacks causing physical harm or critical infrastructure disruption across all business types are considered highly likely to occur within the next two years due to the gap between evolving technical threats and current human response systems.
- Future precautionary norms will likely include deleting voicemail greetings and limiting spoken interaction with unknown callers, driven by the need to mitigate voice replication and video modalities that introduce new threat vectors.
- Traditional annual compliance training is deemed ineffective due to outdated content, whereas AI-driven solutions utilizing deepfakes, personalization, and simulated attacks are predicted to offer superior ROI through experiential learning.
- A continuous arms race involving AI agents for both offense and defense is anticipated over the next decade-plus, with the expectation that protective measures will ultimately prevail, though renewed vigilance is required immediately.