Blockstream Liquid's $320M Hack, Open Source AI Summit Takeaways, Vora Reveal

— Blockstream Liquid suffered a major hack exploiting a consensus algorithm vulnerability that allowed creation of 4,000 LBTC out of thin air; the attacker claimed "white hat" status but retained 600 BTC (15%) after returning 3,400 BTC.

— The vulnerability dated to 2018–2019, and a flawed fix introduced a worse bug; security researchers reported both the initial vulnerability and a correct fix, but dispute arose over whether Blockstream properly reviewed the correction.

— Federation model weaknesses were exposed: signers may trust Blockstream software updates without independent audit, and one functionary (Sideswap) lacked velocity controls and KYC, enabling rapid exit of stolen funds.

— Blockstream announced full 1:1 reimbursement for Liquid users, with the 600 BTC shortfall presumably covered by the company or backers; withdrawals remain disabled pending security restoration.

— The Presidio Bitcoin Open Source AI Summit (150+ attendees, two-thirds non-Bitcoin AI professionals) featured speakers including David Sachs, Ramez Naam, and representatives from Hugging Face and NVIDIA on open vs. centralized AI models.

— An Anthropic employee quit after six weeks and tweeted apocalyptic AI safety warnings in a coordinated media campaign, raising concerns about regulatory capture that could harm open-source AI development.

— Evaluate custodians and counterparties (exchanges, sidechains, Lightning nodes) based on transparency, audits, and track record rather than assuming federation automatically provides security; single reputable custodians may outperform weak federations.

— Monitor regulatory developments around AI safety; open-source model builders and Bitcoin community should engage in policy conversations early to prevent regulatory capture that benefits only large centralized labs.