Book a diagnostic call

Archive for Luca Zollino

The control plane of Bitcoin mining is the risk nobody has priced

Hashrate Redirected by API: What the Luxor Attack Reveals

An AI-assisted attack on Luxor’s fleet management software redirected hashrate from more than a dozen accounts to another pool for less than two hours. The vector wasn’t the protocol or custody, but the control plane: half a dozen unregulated SaaS providers decide where a meaningful share of Bitcoin’s computing power points.

Read more
Washington enforces export controls through one man, and nobody can hedge him

When counterintelligence risk lands inside the index

Nvidia wiped out its 2026 gains in two sessions that erased more than $470 billion in market value, with a committee chairman threatening to subpoena Jensen Huang. The underlying problem isn’t the scandal: it’s that Washington now enforces export controls through one person, and the market has no instrument to hedge that risk.

Read more
Bitcoin's security budget is now whatever AI leaves on the table

The Megawatt Switches Sides: Bitcoin Stops Outbidding AI

Core Scientific will shut down its Bitcoin mining during the fourth quarter and IREN will do the same before year-end: what is being reallocated isn’t machines, it’s grid interconnection contracts. Bitcoin’s security budget is now set as the residue of whatever inference is willing to pay for the same megawatt.

Read more
The attacker no longer breaks the system, it breaks whoever certifies it

When the Attribution Provider Is the Target: The Chainalysis Case

Chainalysis has admitted that an attacker using “offensive agentic techniques” accessed customer data while the firm was tracing the $6.4 billion stolen from Binance. Crypto’s weak link in 2026 isn’t the cryptography — it’s the private layer that produces attribution, blocklists and the tooling everything else is built on.

Read more
The oracle, not the leverage, is the real counterparty in GPU perpetuals

GPU Futures: The Risk Is the Index, Not the Leverage

Injective has listed perpetuals on an index tracking rental prices for five Nvidia GPUs, with up to 10x leverage, just as credit markets slam shut for neoclouds. The problem isn’t the product: it’s that the underlying is a politically administered price reported by a narrow panel, which makes the oracle the real point of failure.

Read more
Banning a model nobody can identify pushes enforcement down into the chip

Washington bans DeepSeek and real control moves to the silicon

On 20 October 2026 the DHS published an order excluding DeepSeek from federal civilian executive branch procurement and systems, including third-party products built on its implementations. With open weights, that exclusion is technically unprovable, which is why enforcement is shifting from the model to the runtime: to the sandbox, the telemetry and the chip that executes it.

Read more
Quantum risk is no longer cryptography, it is a fight over property rights

Who Can Freeze 6.4 Million BTC, and by What Right?

Nine crypto companies launched the Quantum Alliance and failed to agree on the only thing that matters: whether anyone can immobilise 30.2% of Bitcoin’s supply to protect it from a computer that doesn’t exist yet. Quantum risk has stopped being a cryptographic problem and become a question of governance and property rights.

Read more
A decade-old integer overflow exposes who really controls the XRP Ledger

A 2015 bug in XRP Ledger and the end of artisanal auditing

An arithmetic overflow that had been sitting in the XRP Ledger’s payment engine since 2015 made it possible to mint 18 trillion XRP in a single transaction, 184 times the total supply. An AI agent found it, it was patched in three days by bypassing the validator vote, and both of those facts say more about the economics of on-chain security than the spectacular number does.

Read more
The power to freeze arrives long before the procedure to challenge it

Compliance Moves Into the Token: CIP-0113 and the Conduit Case

The Cardano Foundation has shipped a standard that lets issuers freeze, seize and move tokens without the holder’s consent. Conduit’s lawsuit against Tether over $2.76 million in frozen funds shows what happens when that power exists with no procedure to contain it.

Read more
Either the provider knows who paid, or it cannot comply

zkAPI vs. export controls: the collision nobody admits

The Ethereum Foundation has shipped a mainnet system that lets you pay for AI inference without the provider being able to link the request to whoever paid for it. That is the exact opposite of the compliance architecture Washington is building on top of access to frontier models.

Read more
Irreversible money handed to software that cannot tell instruction from content

x402: Irreversible Payments for AI Agents That Still Break

On 25 September, Block joined the x402 Foundation and added Lightning to the payment standard for autonomous agents, alongside Google, Microsoft, AWS and Coinbase. The problem isn’t the rail: it’s that irreversible spending power is being handed to a software layer that this same week exfiltrated files from a government portal and handed its authentication token to a terminal command.

Read more
Washington is rewriting market structure by order, not by law

The US Is Building Its Onchain Stock Market Without Congress

The SEC has granted a temporary five-year exemption allowing tokenized NMS stocks to trade in automated liquidity pools, the CFTC has sent its own rulebook to the White House, and Coinbase wants to list perpetuals on Apple, Tesla and Nvidia. With the Clarity Act dead in the Senate, US market structure is being rewritten through administrative instruments that the next administration can undo with a single signature.

Read more
Four CEOs set the price of compute without signing a single contract

The AI Slowdown Pact Is Private Monetary Policy

Four labs loosely agreed to “pace the frontier” and the market repriced before a single contract existed. What is being set is not a safety standard: it is the price and cadence of the decade’s scarce asset, compute.

Read more
A private attribution layer now decides who is cut off from crypto

Who Decides Which Crypto Address Is Sanctioned

On 9 September 2026 the Treasury sanctioned the Xinbi marketplace, the Secret Service froze $52.8 million in crypto and TRM Labs doubled its valuation to $2 billion. These aren’t three separate stories: they’re one circuit, and its weak point is a private attribution layer that nobody audits.

Read more
When AI capex stops being equity risk and becomes bank credit risk

$29.6bn unsecured: AI capex moves onto bank balance sheets

ByteDance has closed a $29.6 billion syndicated loan with close to thirty banks, with no collateral attached, to fund AI infrastructure outside China. The deal shifts the risk of the AI cycle from venture capital to bank credit, against assets that cannot be pledged or placed wherever the lender might prefer.

Read more
Every dollar into stablecoins buys far less Treasury demand than Washington claims

Eight Cents: The Arithmetic That Dismantles the Stablecoin Story

A paper by Nellie Liang and Brent Neiman puts a number on how much Treasury bill demand each dollar flowing into stablecoins actually creates: between $0.08 and $0.79, depending on where that money comes from. The figure breaks both the crypto argument and the bank lobby’s, and exposes what the GENIUS Act really contains: a lever of financial coercion.

Read more
Could an AI Anchor Be The Next Step In News And Journalism

Could an AI Anchor Be The Next Step In News And Journalism?

A Revolution in News Delivery In an age of rampant digital deception exemplified by “deep fakes,” the introduction of AI avatar news, dubbed “Deep Real,” heralds a contrasting commitment to truth, transparency, and journalistic integrity. Deep Real vs. Deep Fake: A Contrast in Intention Deep fakes, notorious for deception, find their antithesis in Deep Real. Here lies a genuine embodiment of journalism, which, powered by technology, champions truth and integrity over manipulation. Building Trust in an AI-driven Era Fears surrounding the misuse of AI-generated avatars—particularly using images of journalists without consent—have been fanned by popular culture. Episodes like Black Mirror’s “Joan Is Awful” highlight potential dystopian scenarios. Yet, if wielded with responsibility and rigor, AI can amplify transparency in journalism, building trust in ways we’ve yet to fully realize. Beyond Traditional Journalism with AI Deep Real avatars can break conventional news boundaries, allowing for clearer, more globally-reaching reporting. These digital representations don’t undermine human roles. Instead, they empower journalists to delve deeper into investigative and analytical aspects of their profession. Automated processes free human resources to engage more intimately with sources, enhancing the quality and depth of news stories. Ethics and Integrity in AI Journalism Ensuring ethical AI journalism requires clear guidelines and industry standards. Full transparency is essential, making audiences aware when they’re interacting with AI avatars rather than humans. News outlets must maintain unwavering accountability for content, emphasizing rigorous fact-checking and quality control. The Bright Future of Journalism with AI AI heralds not an end, but a renaissance for journalism. It offers the chance to democratize news access, tailor storytelling, and magnify the voices of journalists. Deep Real stands as a testament to the fusion of human creativity and technological progress, driving us towards an era of boundless storytelling potential.Conclusion: Incorporating Deep Real responsibly into journalism paves the way for enhanced connection, inclusivity, and impact. This momentous shift challenges us all—journalists, news outlets, and society—to uphold the age-old principles that have steered journalism, ensuring that as the medium evolves, its core remains unshaken.

Read more
How Could AI Help Make Evaluations More Inclusive

How Could AI Help Make Evaluations More Inclusive?

In the modern educational landscape, exams should be a tool for everyone, ensuring a learner’s aptitude is genuinely reflected. Yet, biases have historically distorted their fairness. Enter AI—a tool with potential to revolutionize inclusive assessment, though not without hurdles. A Real-world Dilemma: Bias in Tests and Exams Exams have traditionally catered to the majority, often overlooking minority groups. From language barriers to socio-economic disparities, biases make assessments less of an accurate measure of knowledge and more of a reflection of privilege. The Global Push for Fairer Assessments The movement towards inclusivity is not just a regional issue. From the UK to the US, institutions recognize the need for change. Practices such as “test-optional” admissions and accessible guidance like Ofqual’s are strides in the right direction. AI’s Role in Reshaping Assessments The rise of AI in learning and assessment cannot be ignored. From AI-created questions to the potential risks of cheating using tools like ChatGPT, the realm of AI-assisted education is vast and still developing. Can AI Foster or Hamper Inclusivity? The bias in AI is a genuine concern. Tools like ChatGPT may inadvertently reflect societal biases, potentially leading to further marginalization in assessments. However, these tools are in their infancy, and their potential for positive change is vast. The Potential of AI for Equitable Assessments As AI matures, there’s hope it will help make exams more unbiased. By identifying potential bias in questions, offering personalized learning, and even aiding in the fight against cheating, AI may well become the great equalizer in education. A Glimpse into the Future of AI and Assessments While challenges persist, AI’s promise lies in its continuous evolution. With the right training and approach, AI has the potential to make assessments more considerate, respectful, and truly reflective of a student’s abilities.

Read more
Educators integrating AI in the classroom

How Are Educators Using AI As An Educational Tool?

In the revered corridors of institutions like Yale, Artificial Intelligence (AI) is becoming an integral part of the academic fabric. The Timeless Question: What is Education’s Purpose? Is school’s role simply to memorize? Or to sculpt minds to tackle real-world challenges? The perception becomes the lens through which educators view AI in classrooms. Two Educators, One Vision Meghan Tocci and Gary Marchant, distinct in their teaching domains, find common ground in their stance on AI. Their consensus? AI’s integration into classrooms isn’t a probability—it’s imminent. The Multifaceted Role of AI in the Curriculum While AI shines in pattern recognition, Tocci and Marchant believe its true potential lies in assisting, not replacing. As Marchant highlights, “Preparing students for real-world law practice means introducing them to AI tools.” Yale’s Pioneering Approach Yale’s courses encourage AI-centric discussions, pushing students to discern AI’s capabilities versus human uniqueness. This balance, Tocci believes, is key to future education.H2: The Inevitable IntegrationDespite some skepticism, both educators are confident that the widespread adoption of AI in classrooms is just around the corner.

Read more
How Will Google’s AI Improve People’s Decision-Making?

How Will Google’s AI Improve People’s Decision-Making?

In an age where artificial intelligence (AI) pervades various facets of our lives, Google’s DeepMind takes a bold step towards harnessing AI’s capabilities to provide life advice. But with great power comes great responsibility – how will this tech giant ensure the safe and beneficial use of such tools? DeepMind’s Ambitious Endeavor: Life Advice Tools Powered by AI As reported by The New York Times, Google’s AI division, DeepMind, is in the process of creating a suite of at least 21 tools geared towards life advice, planning, and tutoring. Recognized for its agility and innovation, DeepMind continues to lead the charge in Google’s AI ventures. The Collaboration with Scale AI and Expert Input In their pursuit of excellence, Google has teamed up with Scale AI, a renowned startup valued at $7.3 billion, specializing in training and validating AI software. The project is backed by a massive talent pool, with over 100 Ph.D. experts diligently working on it. Their meticulous approach involves probing the AI’s capabilities, even in sensitive areas like offering relationship advice or responding to personal dilemmas. The Ethical Implications and Public Concerns However, as AI delves deeper into the personal domain, it raises concerns. Notably, Google’s AI safety experts have highlighted potential risks like “diminished health and well-being” and a possible “loss of agency” for users relying on AI for critical life advice. These concerns aren’t unfounded. AI tools venturing into the realm of therapy or medical advice have faced scrutiny in the past. For instance, the National Eating Disorder Association’s chatbot, Tessa, was suspended due to its provision of potentially harmful advice. Google’s Stance: Treading Carefully in Uncharted Waters DeepMind’s tools, as reported, aren’t designed for therapeutic applications. In fact, Google’s existing Bard chatbot is programmed to offer mental health support resources instead of direct therapeutic counsel. This cautious approach is rooted in the ongoing debate surrounding AI’s role in therapeutic and medical contexts. However, Google remains committed to safe and effective AI deployment. A spokesperson from Google DeepMind mentioned their longstanding collaboration with various partners to assess their products, emphasizing the importance of creating technology that’s both helpful and safe. The Road Ahead: Navigating the Complex Landscape of AI and Personal Advice AI’s role in personal life advice remains a controversial and fascinating frontier. While its potential to provide valuable insights and assistance is undeniable, ensuring its responsible and ethically sound implementation is crucial. As Google DeepMind embarks on this journey, the tech world and its users will keenly observe how AI tools evolve in the context of offering personal advice. In Conclusion Google’s DeepMind’s venture into AI-powered life advice tools presents a thrilling yet challenging endeavor. As technology continues to shape our lives, the intersection of AI and personal life guidance is set to be one of the most intriguing domains to watch.

Read more
Telegram Message us on WhatsApp