Topic
ai
15 stories
NEAR AI breach: SGLang flaw let attacker read user account details and live requests
NEAR AI says an attacker used a remote code execution flaw in the open-source SGLang inference engine to pull account details for about 100 users and access 18 legacy agents before access was contained.
GitBook Reports AI-Agent Breach, Tells Users to Rotate API Tokens
Documentation platform GitBook says an attacker used AI agents to reach some of its infrastructure credentials on November 2, exposing API tokens and integration secrets and prompting urgent key rotations.
Luxor raises $16M to turn AI computing power into something you can trade
Luxor Technology raised $16 million led by CMT Digital, with mining-rig makers Canaan and MicroBT among the backers, to build out hardware brokerage and derivatives for trading AI computing power.
Bitcoin developers rework security practices as AI makes bug-hunting cheap
Bitcoin Core is revising its security disclosure policy as AI lowers the bar for finding software bugs, while a volunteer group called the Bitcoin Red Team hunts AI-assisted vulnerabilities across the ecosystem.
SEC puts tokenization and "super intelligence" on its FY2027 exam list
The SEC's Division of Examinations named tokenization, Treasury clearing reforms, and "super intelligence" among its FY2027 exam priorities, listing them alongside long-running focus areas like adviser conflicts and retail protection.
SEC Sets November 9 Roundtable on "Super Intelligence" and Machine-Readable Data
The SEC will hold a public roundtable on November 9 to discuss machine-understandable data and what it calls "Super Intelligence," gathering market feedback rather than proposing any rule.
Ethereum Foundation ships zkAPI, a pay-for-AI system that hides who's paying
The Ethereum Foundation and the Open Anonymity Project put zkAPI live on Ethereum mainnet, a system that lets users pay for metered AI services using zero-knowledge proofs instead of identity-linked API keys.
Ethereum Foundation ships zkAPI, a zero-knowledge way to pay for AI without an identity attached
The Ethereum Foundation and the Open Anonymity Project launched zkAPI, a system using zero-knowledge proofs to let users pay for metered APIs without linking the payment to their identity or prompts.
SEC charges four entities over $15M in fake "AI trading" crypto scams that forged SEC registration
The SEC charged four offshore-linked entities behind two "AI trading" crypto scams that faked SEC registration and stole at least $15.3 million from hundreds of retail investors, many in the U.S.
CFTC to Launch Public "Frontier Forum" Roundtables, Starting With AI and Agentic Finance
The CFTC's Innovation Task Force will launch a Frontier Forum roundtable series on financial technology, opening October 28, 2026 with a session on AI and "agentic finance.
New research says AI agents aren't really spending money online yet
A Decrypt report points to new research finding that AI agents are not yet meaningfully spending money online, cutting against a heavily marketed "agentic commerce" narrative.
Cango stock drops ~21% after $81.6M quarterly loss as miner shrinks its fleet
Bitcoin miner Cango's stock fell about 21% on September 1 after it reported an $81.6 million second-quarter net loss and roughly halved revenue while shrinking its mining fleet.
World Liberty says WorldClaw is 'independent' — and won't say what, if anything, it owns
World Liberty Financial says it neither owns nor operates Hong Kong AI reseller WorldClaw, but declined to say whether it holds equity, financing, licensing fees or revenue share in the company.
Bitcoin mining difficulty falls below year-ago levels for only the second time ever
Bitcoin's mining difficulty has dropped to 126.23 trillion, 1.1% below where it sat a year ago — only the second annual decline ever, after China's 2021 mining ban.
Ethereum Foundation: AI agents found real protocol bugs, but the triage is where the work is
Ethereum Foundation's Protocol Security team says coordinated AI agents found real bugs in protocol code, including a disclosed libp2p flaw, but that verifying the findings — not generating them — is now the hard part.