When businesses increasingly rely on AI to make decisions, a manipulated dataset can cause more harm than a ransomware attack, says Gerald Beuchelt, Chief Information Security Officer at Acronis. In an interview with TRENDS, he said that in the present day, deepfakes, synthetic identities, and fabricated evidence are blurring the lines between what is real and what is artificial.
In the knowledge economy, trust and data protection are critical. How can cybersecurity evolve to protect innovation itself?
Innovation depends on integrity. It takes only a single corrupted dataset or an unnoticed piece of tampered code for months of work to unravel. We’ve seen organizations become stuck not because they lacked ideas, but because a single breach forced them to stop, clean, verify, and rebuild before they could innovate again.
Most recently, we witnessed the impact IT failures can have on innovation when the services of leading global cloud providers went dark, essentially causing digital services for businesses around the world to grind to a halt.
Without assurance in the integrity and capabilities of underlying digital platforms, businesses won’t have the confidence to innovate. Cybersecurity must therefore evolve from simply blocking threats to actively safeguarding the integrity of the innovation process.
At Acronis, we see this shift happening when organizations validate every critical asset—from code to training data—using cryptographic signatures, treat change histories as immutable records, and build environments where only the right people, at the right moment, can alter something meaningful.
Companies that adopt these best practices are able to restore operations in minutes rather than days. As a result, they are the ones that continue to innovate even when adversity strikes.
How is Acronis approaching the balance between openness for collaboration and protection for security?
Collaboration without guardrails is often where breaches begin. The challenge is enabling people to work freely without exposing the organization to unnecessary risk.
We approach this by treating collaboration as a governed, contextual privilege. Imagine a project team spread across three countries, each using different devices and networks. Traditionally, everyone might be granted broad access until the project ends. In reality, people join and leave, roles change, and external partners come and go.
Our approach continuously evaluates identity, device posture, and location to ensure access extends only as far—and for as long—as necessary.
Protection does not mean locking everything down. Instead, it means embedding security directly into the collaboration flow. We classify data at the moment it is created, and those classifications travel with the data wherever it goes.
The goal is an ecosystem where openness drives progress and trust ensures it does not come at the cost of security.
What forms of cyber risk do you foresee emerging with the rise of AI-generated knowledge?
There are multiple facets to this. One of the most concerning is the quiet corruption of data and models. We have already seen cases where poisoned training data caused AI systems to behave unpredictably because the foundation itself was compromised.
When businesses increasingly rely on AI to make decisions, a manipulated dataset can cause more harm than a ransomware attack.
There is also the growing risk of intellectual property leakage. Developers often feed code or architecture diagrams into generative tools without realizing these interactions may later be inferred through model inversion techniques. Sensitive material can be reconstructed or unintentionally revealed, creating a new pathway for corporate espionage.
On a more human level, customers and employees could be impacted by the collapse of authenticity. Today, deepfakes, synthetic identities, and fabricated evidence are blurring the line between real and artificial. When you cannot trust what you see, the entire digital environment becomes harder to secure and govern.
Mitigating these threats requires more than new tools. It demands traceability. Organizations need confidence that the data used to train models comes from approved sources. They need provenance and watermarking to verify whether content is real. They need large language model interfaces designed with guardrails to prevent leakage, while continuing to practice core security disciplines such as least privilege, strong encryption, and continuous monitoring.
While many of these risks are new, time-tested security best practices still apply.
How can public-private partnerships strengthen the resilience of digital knowledge ecosystems?
Public-private partnerships must focus on practicality rather than posturing. When governments, industry players, and technology providers share threat intelligence in near real time, everyone benefits. A small clinic or logistics company, for example, may not have a dedicated cybersecurity team, but through shared intelligence, it can gain early-warning insights that help protect critical operations.
Joint exercises are equally important. When a crisis hits, it is too late to exchange business cards. Practicing together ensures everyone understands their role in detection, containment, and recovery, dramatically shortening decision-making time during real incidents.
There is also long-term value in agreeing on common guardrails. Standards such as software bills of materials (SBOMs), transparent disclosure practices, and digital provenance help ensure integrity across borders and vendors. These frameworks mean that if one part of the ecosystem is compromised, the entire system does not grind to a halt.
The most transformative investment, however, is in people. Skilled defenders amplify the impact of every tool and policy. When administrators know how to perform rapid forensics, when developers understand secure coding from day one, and when students gain hands-on experience with simulated attacks, the region becomes far more resilient.
Real-world practice builds a kind of muscle memory that no automated system can replicate.
Ultimately, resilient knowledge ecosystems do not emerge from isolated excellence but from collective readiness. Public-private partnerships provide the structure needed to achieve that readiness at scale.
