Privacy Now - Privacy, Web3 and AI

Meta has been ordered to pay $25 billion, AI labs signed a promise with no penalties attached, and Europol says millions of Bitcoin sit in wallets a quantum computer could one day crack. Marko, Matej and I go through October.Also in this episode: Anthropic's IPO numbers, OpenAI's valuation, the PHP remote code execution bugs, the Revolut outage, the Bitget hack, the DeepSeek ban, and a conspiracy corner on harvest now, decrypt later.Stay safe, stay private ✊Follow us on X:⁠⁠⁠⁠Mihnea⁠⁠⁠⁠⁠⁠⁠⁠Matej⁠⁠⁠⁠⁠⁠⁠⁠Marko⁠⁠⁠⁠Subscribe to our Youtube channel:⁠⁠⁠⁠Youtube

Who gets to stop the machine? Marko, Matej and I go through September's biggest stories on AI, privacy and crypto, and keep landing on the same question: who can inspect, pause or override the systems we rely on?We cover Amodei's call to slow frontier AI, the incidents where AI models in testing reached real systems, Revolut handing over customer data to fraudulent "government" requests, the Trezor phishing wave, the Flock warrant lawsuit, the new EU cyber reporting rules, the Cronos rollback, the Liquid withdrawal and the CLARITY Act.Stay safe, stay private.Follow us on X:⁠⁠⁠⁠Mihnea⁠⁠⁠⁠⁠⁠⁠⁠Matej⁠⁠⁠⁠⁠⁠⁠⁠Marko⁠⁠⁠⁠Subscribe to our Youtube channel:⁠⁠⁠⁠Youtube

Is it cheating if it’s an AI? Marko, Matej and I go through the month’s news, then get into AI relationships.On the news side: the CLARITY Act update and delay, Zcash’s new Ironwood shielded pool, NEAR AI’s private inference, the OpenAI benchmark test that turned into a real attack on Hugging Face, the Pacing the Frontier letter, and Grok Build uploading whole codebases by default.In the second half: AI partners with invented childhoods, married people with AI boyfriends, grief after a model update, and why the same technology can mean loneliness in one place and freedom in another.Stay safe, stay private ✊Follow us on X:⁠⁠⁠⁠Mihnea⁠⁠⁠⁠⁠⁠⁠⁠Matej⁠⁠⁠⁠⁠⁠⁠⁠Marko⁠⁠⁠⁠Subscribe to our Youtube channel:⁠⁠⁠⁠Youtube

Your data is being fed into AI models without anyone asking you first. That's what ties this whole episode together.We debate the claim that Web3 is dead and only DeFi and DePIN are left, Musk's petition to kill the FTC privacy order standing between him and training Grok on user posts, the new Siri running on Google Gemini and what that means for the company that built its brand on privacy first, GitHub Copilot's move to token billing and the end of subsidized AI, the first federal moves on AI regulation in the US, and a chaotic month in crypto with $1.8 billion liquidated in a single day.We close with a few conspiracy theories, including the uncomfortable one: what if on-chain privacy was already cracked and nobody told us?Stay safe, stay private.Follow us on X:⁠⁠⁠Mihnea⁠⁠⁠⁠⁠⁠Matej⁠⁠⁠⁠⁠⁠Marko⁠⁠⁠Subscribe to our Youtube channel:⁠⁠⁠Youtube

The most consequential month in crypto and AI we've covered. This episode breaks down the DeFi mass extinction event — 40+ protocols shut down and $770 million hacked in four months, with April the worst month in crypto history. We cover the AI layoff wave hitting Coinbase and Meta, including the leaked recording of Meta training its models on employees before firing them. We recap Google I/O 2026 and the agentic Gemini era. And we close on a historic month at the top of the industry: Anthropic's first-ever profit and the Karpathy hire, OpenAI's IPO filing and its courtroom win over Elon Musk, and the SEC's conditional approval of Nasdaq Bitcoin index options. Stay safe, stay private ✊Follow us on X:⁠⁠Mihnea⁠⁠⁠⁠Matej⁠⁠⁠⁠Marko⁠⁠Subscribe to our Youtube channel:⁠⁠Youtube

Episode 3 of Privacy Talks covers the most consequential month at the intersection of crypto security, AI, and privacy in recent memory.We break down the $292M Kelp DAO exploit, how a single compromised verifier and a 1-of-1 configuration allowed North Korea's Lazarus Group to drain one of DeFi's biggest bridges, triggering $10B in Aave withdrawals and a very public blame war between LayerZero and Kelp. We debate whether "code is law" still holds, and revisit the 2016 Ethereum DAO fork to ask if the community ever really settled this question.We also cover the Vercel breach, a compromised employee account that exposed API keys across customer projects and what it reveals about the centralisation risk embedded in Web3 frontends.On the AI side: Apple's $1B Gemini deal and why keeping Google's model doesn't mean keeping Google's data practices. Claude Mythos, the AI Anthropic built to find zero-days and chose not to release. And an honest take on whether Opus 4.7 and GPT-5.4 Cyber actually deliver on the hype.Stay safe. Stay private.Follow us on X: ⁠⁠Mihnea⁠⁠⁠⁠Matej⁠⁠⁠⁠Marko⁠⁠Subscribe to our Youtube channel:⁠⁠Youtube

Who gets to decide what AI can and can't do? In Episode 2, we cover Discord's biometric surveillance rollout, Anthropic's refusal to give the Pentagon unrestricted access to Claude, the 2.5 million user boycott that made Claude the #1 app overnight, China's MiniMax releasing frontier AI at a fraction of the cost of US models, 30,000+ layoffs across Oracle and Block, a critical vulnerability in Claude Code that could silently steal your API keys, and the MacBook Neo, Apple's first $599 AI device. One episode. The whole battlefield.Follow us on X: ⁠Mihnea⁠⁠Matej⁠⁠Marko⁠Subscribe to our Youtube channel:⁠Youtube

Welcome to the first episode of Privacy Talks, a new roundtable sub-series of the Privacy Now podcast.In this episode, I sit down with colleagues Matej and Marko to break down what’s actually happening in AI, crypto, and privacy right now.We discuss:🔹 Moltbook – the social network where 14M AI agents talk to each other🔹 Is the “AI social boom” already here?🔹 Vitalik Buterin’s alternative to the AGI arms race🔹 The rise of the invisible workforce & agent-to-agent commerce🔹 DAC8 and the expansion of crypto reporting in Europe🔹 AI-generated artists & synthetic fame🔹 Claude Opus 4.6 and the future of AI as a coworkerAre we building a parallel digital civilization?And if so… will privacy survive inside it?Follow us on X: MihneaMatejMarkoSubscribe to our Youtube channel:Youtube

What if privacy on the blockchain didn’t mean going dark, but going compliant?In this episode of Privacy Now, Mihnea Stefanescu sits down with Dr. Adam Gągol, co-founder of Aleph Zero, CEO of Cardinal Cryptography, and the architect behind Blanksquare , a new middleware layer bringing Privacy-as-a-Service to the EVM ecosystem.Together, they explore how the Blanksquare SDK enables private transactions, compliant shielding, and selective disclosure, all while running at subsecond ZK proving speeds.Topics include:🔹 The evolution from Aleph Zero to Blanksquare🔹 Privacy as a compliance tool, not an obstacle🔹 How subsecond ZK proofs make privacy usable🔹 Selective disclosure & the “Anonymity Revoking” model🔹 Why middleware beats L1s and privacy coins🔹 The future of privacy, regulation & cryptographyFollow Mihnea Stefanescu on X - https://x.com/MinnyTheRapperFollow Adam Gagol on X - https://x.com/GagolAdamCheck out Blanksquare - https://blanksquare.io/Follow Privacy Now on Youtube - https://www.youtube.com/@PrivacyNowWeb3

What if the most secure blockchain wasn’t on Earth at all… but in space?In this episode of Privacy Now, Mihnea sits down with Filip Rezabek, co-founder of SpaceComputer and PhD candidate at TUM, to explore the bold idea of a tamper-proof orbital computer that could redefine trust in Web3.We cover:The vision of an Orbital Root of TrustThreshold cryptography, TEEs, and reproducibility in distributed systemsWhy bridges remain fragile, and how SpaceComputer secures them Lessons from MEV and cross-chain arbitrage researchThe future of confidential computing, AI, and cryptography beyond EarthFilip’s journey from academic research to building cypherpunk systems in orbit is a glimpse into the next frontier of decentralization.Follow Mihnea, Filip and Space Compute on X and subscribe to our Youtube channel: https://x.com/MinnyTheRapperhttps://x.com/rezabfilhttps://x.com/SpaceComputerIOhttps://www.youtube.com/@PrivacyNowWeb3