(Kitco News) – As the digital revolution powers full steam ahead thanks to the rise of blockchain technology and artificial intelligence (AI), concerns around privacy and surveillance continue to mount as recent developments point to the Orwellian future that many have long feared.
To get a boots-on-the-ground perspective on Oracle’s recent advancements in using AI to analyze surveillance data and explore what can be done to protect individual privacy in the world of big data and constant online surveillance, Kitco Crypto spoke with Yannik Schrade, co-founder and CEO of Arcium, the first parallelized confidential computing network.
The conversation started with Schrade citing the recent comments Oracle founder Larry Ellison made about the technology, where he boasted about the capabilities of Oracle AI to analyze surveillance data.
“Larry Ellison's recent comments affirmed, in many ways, that our world has already overtaken George Orwell’s dystopian vision,” Schrade said. “In his words, AI will usher in an era of pervasive monitoring and ensure that ‘citizens will be on their best behavior.’”
“Generative AI has made the idea of individual privacy seem redundant,” he underscored. “Tools like Chat GPT and Gemini are trained on massive datasets often scraped without consent, containing personal information ranging from public records to social media images. Every prompt, every question asked to those models contains sensitive information that is being jeopardized.”
“The uncontrolled data usage makes it impossible for individuals to manage their privacy,” he declared. “However, the onus should never be placed on the individual. We must demand more from our builders.”
“The pace of AI development highlights the need for a new approach to protecting personal information and greater accountability in how AI systems are trained and deployed,” Schrade stressed. “AI is advancing, but encryption is stuck in the past. With Arcium, we’re fixing this. Our network provides a trustless environment where data remains encrypted during computation – no decryption is needed. Malicious actors have no entry point, enabling users and developers with secure AI training and inferencing.”
Regarding which areas of day-to-day life are most at risk from this intrusion, Schrade noted that “Every surveillance initiative thinks that they're making their community safer. Yet, corporate surveillance sold for consumer purposes can just as easily be used as a tool for tyranny – and history shows us that it will.”
“The societal implications of private surveillance partnerships like Google’s Nest and Amazon’s Ring are staggering,” he warned. “With upcoming wearable technology like physical AI companion devices or smart glasses – like Meta’s Orion glasses – surveillance is becoming more pervasive.”
“There have been pilots of using open-source facial recognition models trained on social media profiles to identify any person viewed through wearable smart glasses like Meta’s Orion glasses,” he highlighted. “It’s easy to see why this is extremely dangerous. Those are highly problematic developments, and we require adequate responses as societies with their increasing adoption.”
Schrade said that one of the major issues with encryption and privacy technology “has been framing it solely as a political matter. For too long, we’ve been sold a false choice between privacy and security due to it being framed purely as a political narrative.”
The solution, Schrade argued, is to focus on the use of decentralized AI.
“Decentralized AI uses decentralized networks rather than centralized technologies to improve transparency, privacy, and security in AI development,” he explained. “This shift promises a more democratic, resilient, and trustworthy AI ecosystem.”
“However, traditional blockchains (as the most modern form of distributed, decentralized systems) are inherently transparent, which, while beneficial for transparency and trust, poses a significant limitation for AI development that requires data confidentiality,” he noted. “This is where decentralized confidential computing (DeCC) comes into play.”
“DeCC combines two powerful concepts: decentralization, where data is distributed across multiple locations, and confidentiality, which keeps data private and protected from unauthorized access,” Schrade said. “There are several decentralized confidential computing techniques being tested at the moment, including Zero-Knowledge Proofs (ZKPs), Fully Homomorphic Encryption (FHE), and Multi-Party Computation (MPC).”
“All of these methods are essentially trying to do the same thing – verify essential information without disclosing sensitive information from either party,” he explained. “MPC has emerged as a frontrunner for DeCC, enabling verifiable, transparent settlement and selective disclosure with the greatest computational power and efficiency.”
Acknowledging that it’s a stretch to expect companies to focus on creating decentralized AI anytime soon, Schrade underscored the importance of developing ethical AI training models while the development of decentralized AI applications unfolds.
“I'm not against data collection. I love the idea of improving my life using data. But people should have the right to choose for themselves how their data is used and who gets access to it,” he said. “Instead, most have no idea what's going on and no control over where their data ends up. Confidentiality is not about restricting the potential for data exploration; it’s about opening up entirely new use cases and ways to utilize data in a safe and trustless way, giving back ownership and creating new economic incentives.”
“Imagine training models on distributed sensitive data and running inference with zero risk of leaks, all without sharing data with third parties,” he posited. “There’s no need for trust, yet new, unimaginable potential insights emerge.”
“We have the technology to eliminate intermediaries entirely,” Schrade emphasized. “You and I can train a model or perform data science on isolated, sensitive data without sharing any of it – while still gaining insights. The same applies to predictions or inference without involving any party that can see the input data or output result.”
He said this worldview has influenced the development of the Arcuim Network.
“Multi-Party Computation eXecution Environments (MXEs) are the backbone of the Arcium Network, combining MPCs, semi-homomorphic encryption, zero-knowledge proofs, and more to enable encrypted computations,” he explained. “MXEs are highly configurable, allowing users to customize everything from trust assumptions to hardware power. They facilitate the secure pooling of encrypted data from multiple sources, enabling collaborative AI model training without exposing individual datasets.”
“This approach expands the scope and diversity of data available for AI development while preserving each contributor's privacy,” Schrade said. “Decentralized confidential computing in AI can transform the future, and we are building it!”
As for how this design will help combat the dangers of centralized surveillance systems, he noted that “Arcium’s decentralized confidential computing network enables a world where we can analyze, process, and learn from information without exposing it.”
“Our vision is to redefine every digital interaction by making the web run entirely on encryption,” Schrade said. “Arcium leverages cryptographic techniques like secure MPCs to allow computations over encrypted data without exposing any sensitive information. This ensures that personal data remains confidential and cannot be misused.”
“By decentralizing control and encrypting data throughout the entire process, Arcium’s network prevents the overreach of centralized surveillance, protecting data privacy and fostering a more secure digital environment,” he added. “This not only empowers current systems but opens up completely new design spaces capable of transforming the fabric of the internet.”
As for the motivation behind his quest to protect individual privacy and reshape how AI is used for surveillance, Schrade cited the book 1984, which fundamentally altered his worldview when he was young and set him on a course to champion the rights of everyday people.
“Reading 1984 as a young child had a lasting impact on my perception of how the world truly operates,” he said. “It opened my eyes to the importance of privacy and freedom, concepts that became especially significant to me as I began learning programming and studying law. It sparked a fascination with encryption and set me on a path toward building something far more ambitious: a global supercomputer for a hyper-encrypted web.”
