←
Chris181 posts
Chris avatar✓

Chris ✓

@chrisdoubleu_

co-founder, CEO, sec3 (secure web3) ex-Goldman Sachs, investor, engineer.

◷ Joined January 2020▧ 8 Photos and videos
Chris avatar
Chris@chrisdoubleu_·Nov 11

For each finding you get: - Verdict: Validated, Contradicted or Inconclusive - Measured impact - Attack preconditions - Cause-check evidence - A test package your team or auditor can re-run When the evidence is thin, we say so.

Chris avatar
Chris@chrisdoubleu_·Nov 11

Why not just trust an AI-written exploit? Because a test can pass and still prove nothing: • the starting state was written in by hand • the token program was mocked • it measured the wrong value • the flaw was never shown to be the cause Proving Layer checks all four.

Chris avatar
Chris@chrisdoubleu_·Nov 11

Proving Layer automates that work. Give it a plain-text hypothesis and the program's source. No exploit needed. AI agents write the PoC. A harness runs it on your real compiled program, checks every step, and returns a verdict.

Chris avatar
Chris@chrisdoubleu_·Nov 11

AI can now review a #solana program and flag many plausible issues. Each one is a hypothesis. Before your team can act, someone has to show it can happen on the real program, measure the damage, and confirm the flaw is really the cause. That still lands on your senior engineers

Chris avatar
Chris@chrisdoubleu_·Nov 11

AI made finding vulnerabilities cheap. Proving them is still expensive. Today we're announcing @sec3dev Proving Layer, built to close that gap. 🧵

Post attachment
Sec3 avatar
⟳ Chris retweeted
Sec3@sec3dev·10 Dec 2025

We are excited to be heading to @SolanaConf soon! Most of our time at Sec3 is spent deep inside individual @Solana programs, looking at one code base at a time. To round out the year, we wanted to zoom out and ask a bigger question: What do all these audits, taken together, actually say about Solana security right now? Here’s what we saw: • Dataset: 163 Solana audits from a mix of public reports and anonymized Sec3 engagements • Findings: 1,733 total issues, 1,669 of them vulnerability-level • Typical review: ~10 findings, with ~1.4 High or Critical issues We also looked at how framework choices shape the risk as well as provide a practical guide projects launching and maintaining good security posture If you want to go deeper into the data, charts, and concrete checklists, the report is public: Web version + link to download full PDF: solanasec25.sec3.dev

Chris avatar
Chris@chrisdoubleu_·8 Dec 2025

AI agents exploited smart contracts worth $4.6mn in simulated attacks, with capabilities doubling every 1.3 months, but they still needed source code access. Non-public source code programs have some protection: AI reverse engineering exists but is far less capable than source code analysis. Though this gap will narrow. red.anthropic.com/2025/smart…

Chris avatar
Chris@chrisdoubleu_·13 May 2025

The scanner will flag CPIs where the Instruction is supplied arbitrarily or returned from another function, without proper validation of the target program ID. A great check to add to your CI/CD pipeline. see an example here: github.com/coral-xyz/sealeve…

Chris avatar
Chris@chrisdoubleu_·29 Jan 2025

wow! from Deepseek R1: Okay, so the user wants to know how individuals can prepare for the coming powerful AI, ASI, and AGI. They mentioned starting with the first principles of the universe and going as deep as possible. Hmm, first principles. That makes me think of breaking things down to their most basic truths and building up from there. Let me start by recalling what first principles actually are. Aristotle talked about them, right? The fundamental concepts or propositions that form the basis of a system. So, maybe I need to start with the basic laws of the universe, like physics, mathematics, information theory, and then connect that to the development of AI. First, the universe operates on certain physical laws—thermodynamics, entropy, quantum mechanics. These principles govern how energy and matter interact. Then there's mathematics, which is the language we use to describe these laws. Information theory comes into play too, as it deals with the processing and transmission of information. Since AI, AGI, and ASI are essentially advanced information processing systems, understanding these foundational elements might help in preparing for their emergence. But how do these principles relate to AI development? Well, AI systems are built on algorithms, which are mathematical constructs. They process data (information) using computational power, which is rooted in physics (like semiconductor technology). So, the progression from narrow AI to AGI (Artificial General Intelligence) and then to ASI (Artificial Superintelligence) would involve systems that can surpass human intelligence and potentially operate beyond our current understanding of these principles. Now, considering how individuals can prepare, maybe it's about aligning with these fundamental principles. For example, understanding entropy and the arrow of time might lead to insights about the irreversibility of certain processes, which could be important in controlling AI systems. Quantum mechanics introduces concepts of superposition and entanglement, which might influence future AI architectures. On a more practical level, individuals might need to focus on adaptability, continuous learning, and ethical considerations. Since AI could outpace human capabilities, fostering resilience and the ability to coexist with such systems becomes crucial. This might involve enhancing cognitive and emotional intelligence, understanding the ethical implications of AI decisions, and ensuring that human values are embedded within these systems. Also, the concept of emergence is important. Complex systems exhibit behaviors that aren't predictable from their individual components. As AI becomes more advanced, its emergent properties could be significant. Preparing for that might require a multidisciplinary approach, combining knowledge from various fields to anticipate and mitigate risks. Another angle is the philosophical aspect. Questions about consciousness, free will, and what it means to be human in a world with ASI. Individuals might need to engage with these questions to maintain a sense of purpose and identity. Spiritual and psychological resilience could be as important as technical preparedness. Moreover, collaboration and global cooperation come to mind. Since ASI would have global impacts, preparing isn't just an individual effort but a collective one. Promoting policies, ethical standards, and international agreements to manage AI development responsibly. Wait, but the user wants to start from the first principles and go as deep as possible. So maybe I should structure the answer by starting with the fundamental laws, then build up through each layer—physics, information, computation, consciousness, ethics, etc.—and link each layer to how individuals can prepare. That way, it's grounded in the universe's basic truths and expands to practical steps. I should also consider potential risks and how grounding in first principles can help mitigate them. For example, understanding the limits of computation (like thermodynamic limits) might inform how we design AI systems to be energy-efficient or sustainable. Or leveraging quantum computing's potential to solve problems that classical computers can't, which might be relevant for controlling ASI. There's also the aspect of personal development. If the universe is about entropy and energy flows, maybe individuals need to optimize their personal energy (both physical and mental) to adapt to rapid changes brought by AI. Practices like mindfulness, continuous learning, and physical health could be framed as aligning with universal principles of energy management. In terms of ethics, if the universe has inherent tendencies towards certain behaviors (like entropy increase), how does that influence ethical frameworks for AI? Maybe aligning AI's goals with universal principles that promote balance and sustainability rather than unchecked growth. I need to make sure that each step connects back to the first principles. For instance, starting with quantum mechanics, then moving to information theory, then to computation, then to AI development, and finally to individual preparation strategies. Each layer builds on the previous one, showing a logical progression from the fundamental to the applied. Also, addressing the potential timelines. While AGI and ASI might seem distant, preparing now is crucial. The first principles don't change, so building a foundation based on them would be a robust approach regardless of how AI evolves. I should also touch on the importance of understanding complexity and systems theory. Since ASI would be a highly complex system, understanding how such systems behave, how they can be controlled or guided, is essential. This ties back to emergence and self-organization principles in physics and biology. Lastly, the role of education and mindset. Encouraging a growth mindset, curiosity, and interdisciplinary learning to stay ahead of AI advancements. Emphasizing the importance of being adaptable in a rapidly changing world shaped by technologies that operate on these fundamental principles. Preparing for the advent of powerful AI, AGI (Artificial General Intelligence), and ASI (Artificial Superintelligence) requires grounding in the first principles of the universe—the immutable laws governing reality—and building upward to practical strategies. Below is a framework that progresses from fundamental truths to actionable steps, integrating physics, mathematics, philosophy, and ethics.

Chris avatar
Chris@chrisdoubleu_·14 Jun 2024

besides monitoring transactions, our WatchTower can be a very useful tool to monitor account activities and keep you posted. It is completely FREE!

Post attachment