Rethinking trust when seeing and hearing are no longer believing

1. In your work with UncovAI, how have you seen generative AI fundamentally change the risk landscape for CISOs, particularly when it comes to trust in digital evidence and executive decision-making?

After earning a Ph.D. in applied mathematics and NLP, followed by research in the United States on social network analysis and disinformation in sensitive contexts, I observed the rapid and widespread emergence of AI-generated content. It was in this context that I founded UncovAI, a solution that detects synthetic content (text, audio, images, video) to ensure the integrity of information, both for citizens via our online platform and for businesses via integrations (API, on-premise, Microsoft Teams, Zoom, WhatsApp, Telegram).

Today, with the widespread adoption of GenAI tools, the risk no longer concerns only CISOs but all users: as the use of AI increases, the risk of manipulation and fraud grows exponentially, making accessible detection solutions that are integrated into daily operations indispensable.

Generative AI has fundamentally broken the assumption that digital content can be trusted by default. CISOs now face a shift from securing systems to validating reality itself, as synthetic content can influence executive decisions, fraud approvals, or crisis responses in real time.

2. Deepfakes are often discussed as a media problem but, based on your experience, what's the most underestimated scenario where synthetic content could directly impact a company's leadership or financial integrity?

The most underestimated risk is targeted executive impersonation. While synthetic content is often associated with phishing emails or manipulated images, attacks are becoming far more sophisticated and increasingly rely on synthetic voice and real-time interactions.

In one recent case, a large financial institution was targeted by a voice-based impersonation of a senior executive, creating urgency around a confidential transaction. The request appeared legitimate and aligned with ongoing operations, nearly leading to a significant financial transfer before being flagged. These attacks bypass traditional security controls by exploiting human trust rather than technical vulnerabilities, making them particularly dangerous at the leadership level.

3. You've worked on disinformation and manipulation in sensitive environments. Are we realistically keeping pace with AI-generated deception, or are CISOs already operating at a structural disadvantage?

CISOs are currently at a structural disadvantage, generative AI evolves faster than detection capabilities, and attackers only need one successful deception. Defence requires constant adaptation, while trust erosion is happening faster than organisations can respond.

Based on my experience working on disinformation and social media analysis in sensitive environments in the US, I’ve seen how quickly manipulation techniques evolve.

Today, generative AI accelerates this dramatically. At UncovAI, we are increasingly contacted by banks and large organisations facing deepfake impersonation attempts, particularly involving synthetic voice and executive fraud scenarios.

CISOs are now at a structural disadvantage: generative AI evolves faster than detection capabilities, and attackers only need one successful deception. Deepfake voice attacks now cost companies an average of around $600,000 per incident, highlighting the scale of the risk. As a result, defense requires continuous learning, testing, and adaptation, as trust erosion is happening faster than organizations can respond.

4. In a world where even experts struggle to distinguish real from synthetic, what new skills or cultural shifts should security leaders foster within their organizations to preserve trust internally and externally?

Security leaders must foster a "zero-trust for content" culture, where verification becomes standard practice. This includes training teams to question digital evidence, implementing verification workflows, and combining human judgment with AI-based detection tools.

For example, one organization using our WhatsApp bot was able to detect a deepfake audio impersonation attempt in real time, successfully preventing a fraudulent financial transfer.

5. Will tomorrow's organisations need a 'Chief Authenticity Officer' to defend against AI-generated deception?

Short answer is ‘yes’. I think organisations will likely need a Chief Authenticity Officer, responsible for ensuring the integrity of digital content, overseeing detection of synthetic media, and defining trust frameworks across communications, decision-making, and customer interactions.

While solutions like UncovAI provide scalable and real-time detection, a human-in-the-loop remains essential to interpret complex cases, validate critical decisions, and oversee how algorithms behave in evolving contexts. This role would ensure the right balance between AI-driven detection and human judgment.