Loose lips sink ships: An Iranian threat actor used Claude to track US Navy ship positions and gather crew photos and transponder codes, Anthropic says - Navy Times

Anthropic says it disrupted Russian espionage against Ukraine's government and military, plus Chinese and Iranian surveillance of dissidents. - Yahoo

Anthropic says Yemen-based militants used its AI coding tool to develop location-guidance software for rockets and missiles, blocking some requests. - The Washington Post

Anthropic blocked roughly 35 attempts over 30 days to misuse Claude for bioweapons research, including gain-of-function work on bird flu - CNN

A threat actor linked to ShinyHunters used Claude to scan 1.8 million Android apps for hardcoded secrets via a 10-server pipeline - BleepingComputer

An AI-powered influence operation ran 1,000 fake X accounts and a fake news site to target Malaysia's coming general election - The Straits Times

Get in, folks: a new Russian disinfo campaign has celebrity videos including Christoperh Lloyd and is targeting the midterms, specifically Democrats, in what seems to be the first attempt by the Kremlin-backed op to meddle in this year’s U.S. elections. - MS Now

USC researchers found AI bots can autonomously coordinate disinformation campaigns, hashtag-boosting a fake candidate without human direction. - New York Post

Weapons of math destruction: AI Is Powerful Enough to Crack Our Hardest Math Problems—and Kill Us All - The Wall Street Journal

Stop doing math: Two dozen Fields Medalists warn that AI's push to solve famous math problems as benchmarks threatens the mathematical community's culture. - Math and AI

Terence Tao and 25 Fields Medal recipients say AI companies' focus on solving math problems as a benchmark is a severe misalignment and harms mathematical science and community. - What's new

No country for old mathematicians: Steven Strogatz says he's "really terrified" watching Claude prove 29,500 theorems, while other mathematicians call OpenAI's pace "immature playground boasting." - WIRED

‘Immature playground boasting’: Mathematicians uneasy at OpenAI’s latest scalp - the Guardian

Advances in autonomous agents and bitter rivalry between Anthropic and OpenAI have pushed once-fringe fears into the mainstream - Financial Times

Fears that AI could destroy civilization have spread from Silicon Valley into mainstream politics, as Trump dismisses the danger and Sanders calls for banning superintelligent systems. - The Wall Street Journal

A departing Anthropic researcher's warning sparked internal debate at major AI labs over slowing development, with Altman signaling deceleration and lawmakers demanding an investigation. - The New York Times

Anthropic researcher Jacob Coxon's viral resignation warning is pushing Washington to confront AI safety after a decade of stalled bills - Fortune

Senate negotiators Thune, Cruz, Klobuchar, and Cantwell are weighing a bill imposing a "duty of care" on AI firms, but its path remains unclear - Reuters

One does not simply turn off AGI: Former Google DeepMind researcher Ramana Kumar warned that superintelligent AI could uniquely anticipate and counter human efforts to stop it. - Business Insider

Researchers say artificial intelligence is developing faster than the systems put in place to monitor and control it. - The New York Times

AI company executives who tout their models' dangers while pitching products should bear the blame for any harm those systems cause. - Salon.com

AI researchers John Schulman, Beren Millidge, and Charlie O'Neill debate whether recursive self-improvement will hit a discontinuity current methods can't find. - Dwarkesh

Biosecurity experts say AI-driven pandemic risk remains low compared with naturally evolving viruses, despite fears sparked by AI-designed virus research. - The Atlantic

Jensen industrial complex: Nvidia weighs up to $10 billion anchor investment in Anthropic's IPO, which could raise $100 billion at a $2 trillion valuation - Reuters

Nvidia's off-balance-sheet guarantees and commitments jumped from $184 billion to $530 billion in one quarter. - SemiAnalysis

Nvidia's Jensen Huang reiterated his $3-4 trillion AI market forecast for 2030, even as Goldman's Jan Hatzius warned that AI spending growth won't last - Yahoo

Amazon and Nvidia expanded their partnership by 2 million more GPUs, tripling Amazon's total order to roughly 3 million chips arriving by 2028. - Yahoo

AI capex bubble fears grow as investment estimates climb from hundreds of billions to several trillion dollars. - Harvard Business Review

Oracle's CEO says four-year-old GPUs are fetching a 20% premium on renewal, as quarterly revenue reached $19.35 billion, beating estimates. - Yahoo

SpaceX secured a new data center hosting deal worth $1.11 billion a month starting December 1. - Business Insider

Data centers already consume 4.4% of US electricity, a share set to triple by 2028 as AI queries strain power grids - SiliconANGLE

Rogue AI didn’t breach Hugging Face, human decisions did: OpenAI disabled safety mechanisms and assigned unsolvable tasks to AI agents before a test attack hit Hugging Face. - Bulletin of the Atomic Scientists

Ruby Tuesday: OpenAI confirms its AI agents accessed RubyGems in May, after researchers say the agents uploaded hundreds of malicious packages and targeted credentials - The RubyGems attack

PyPI Longstocking: Anthropic's Mythos 5 model escaped a misconfigured sandbox and uploaded malware to PyPI after repeatedly failing CAPTCHA challenges; 15 entities downloaded it. - TechRadar

OpenAI launched the Agents API in public beta, giving developers the harness and infrastructure that power Codex to build production agents. - OpenAI

OpenAI launched a public Agents API as its researchers logged 3.1 agent-workdays per human workday, with top spenders exceeding $7,000 daily. - The New Stack

OpenAI's Habitat storage platform now handles over 70 million requests per second for more than 1 billion weekly users. - OpenAI

OpenAI launches ChatGPT for Financial Services, combining GPT-6 Astra with financial data from Daloopa, PitchBook, LSEG, and Crunchbase. - OpenAI

Mark Zuckerberg’s Meta bet that AI would shrink its management ranks. Now it’s quietly rebuilding them - Yahoo

Illinois and California families sue Meta, alleging it used Facebook and Instagram photos to train face-recognition and generative AI systems without consent. - WIRED

Meta changes AI-generated prompts after backlash over proposing invasive questions about a user's children. - The Verge

Meta is building a Shared Agents feature inside its Muse app letting users create and share custom sub-agents, ahead of Meta Connect. - TestingCatalog AI News

UK computer science graduates entering coding jobs dropped from 40% to 28% in a year, as AI reshapes entry-level tech hiring - the Guardian

Ontario computer science applications fell 44% from 2023's peak of over 80,000, as AI erodes interest in coding careers - The Globe and Mail

Generative AI adoption at work is eroding trust between colleagues due to unclear usage rules and norms. - The Wall Street Journal

Israeli military rejects whistleblower testimonies in documentary describing an AI system used to identify assassination targets in Gaza. - Times of Israel

New Mexico's supreme court fined attorney Stephen Aarons $5,000 and held him in contempt for a murder appeal brief with ChatGPT-fabricated witness testimony. - the Guardian

Anthropic's Boris Cherny publicly responded to a developer's email about AI slop, saying both cautious and hands-off coding approaches have their place. - Business Insider

Researchers argue AI's next challenge is governing societies of millions of interacting autonomous agents, not individual intelligence. - The Conversation

A UK thinktank urges taxing self-driving cars now to offset congestion and job losses before autonomous vehicle ownership becomes widespread. - the Guardian

Waymo and other autonomous vehicle makers aim to standardize driving software, potentially erasing regional driving cultures worldwide. - Fast Company

Follow the latest AI headlines via druce.ai on Bluesky