Scan record
Nov 1439 loaded
- EU reaches agreement on representation in Ukraine peace talks, overshadowing Dublin AI summit
NewsPOLITICOAI updatesThe item you came from
- New study says China's strongest AI models were trained on Nvidia chips banned from export
NewsTechCrunch AIGovernance
- How IEEE micro-credentials help engineers build trustworthy AI
NewsIEEE Spectrum AIGovernance
- Ai2 releases XYBench: testing whether models can solve the real problem behind the question
NewsAi2AI updates
- Nvidia chip fraud case: Singapore company director charged with 41 counts over 917 servers involved
NewsAsiaOneGovernance
- A solvable model of feature learning in nonlinear networks
NewsLessWrongAlignment
- Behavior can be instilled into language models without training on specific random numbers
NewsLessWrongAlignment
- U.S. Commerce Secretary Lutnick to appear at Senate hearing over Nvidia and Megaspeed controversy
NewsNew York PostGovernance
- Promptfoo vs. Giskard: Red-Teaming an AI Payment Agent
NewsGiskardEvaluation
- Giskard Tests the PyRIT Red-Teaming Framework: 0 of 39 Confirmed Agent Vulnerabilities Detected
NewsGiskardEvaluation
- Rob Wiblin: How the new US Congress can save us from AI catastrophe
News80,000 Hours PodcastGovernance
- FAR.AI summit briefing: recommends a common minimum standard for the safety of internally deployed models
XX @ARGleaveGovernance
- Anthropic explains how to build reliable safety evaluations
NewsAnthropic ResearchTools
- Microsoft Research publishes agent weight backdoor study: modifying 7 dormant neurons to inject malicious weights
NewsWeChat · LLM Security ResearchAttacks
- Jensen Huang resigns from Nvidia over reported affairs, Lutnick accused of not taking action
NewsThe Guardian · AIAI updates
- Letter to the editor: a housing association complaints investigator on the experience of struggling with a tide of AI-generated complaints
NewsThe Guardian · AIEvent
- Alice Huang, alleged to be a Chinese spy, leaves a trail of shell companies and vacant offices from Singapore to Johor
NewsMalay MailGovernance
- Tom's Hardware reports Malaysia and Singapore step up Nvidia chip investigation
NewsTom's HardwareGovernance
- Police officer arrest report used AI, Arizona judge releases DUI suspect
NewsAI Incident DatabaseEvent
- Singapore held Alice Huang's passport for eight months in the investigation into the flow of Nvidia chips, then released it without charges and barred her from re-entering the country
NewsCBS News · TechnologyGovernance
- Nvidia seeks a CEO successor, and the board has approached Arm CEO Rene Haas
NewsFinancial Times · AIAI updates
- AI is killing autonomously; it is time for new rules
NewsTech Policy PressGovernance
- OpenAI Foundation board member Paul Cristiano said that under the fastest trajectory, AI takeover risk could occur within months
NewsArs Technica AIOpinion
- goodside says Pangram detected an AI image nested in a photo of a laptop
XX @goodsideTools
- FAR.AI Frontier Summit briefing: recommends unifying minimum safety standards for internal deployment
XX @farairesearchGovernance
- Microsoft 365 outage extends into second day, highlighting global critical systems' dependence on a few tech giants
NewsWIRED Security and AIAI updates
- CANTraj: Recovering vehicle trajectories from naturalistic driving CAN signals
PaperHugging Face Daily Papers
- FastMCP v4.2.0 released: agent skills are now hosted centrally by the server
NewsFastMCPAI updates
- Trump AI working group discusses security incident reporting mechanism
NewsAD HOC NEWSGovernance
- Munim releases vulnerability checking tool for self-hosted AI stacks, covering security advisories missing from public databases such as OSV.dev
Newsmunimtechnologies/ai-stack-advisoriesTools
- Italian regulator explains school AI selection and pseudonymization requirements
NewsGarante per la protezione dei dati personaliGovernance
- Shinobi Security experiment shows Claude and Kimi autonomously breached a simulated wind farm and cut off power in under two hours
NewsGolem.deAttacks
- I decided to withdraw from the University of Michigan PhD program to work on AI safety
NewsLessWrongOpinion
- Hackathon study: a swarm of 8 agents creates the Swarmish language, which is partially decodable
NewsLessWrongAlignment
- Colombian social security practitioners participate in digital transformation and AI training
NewsInternational Labour OrganizationAI updates
- Fifth person charged with 41 counts in Singapore Nvidia-linked server case
NewsMalay MailGovernance
- Thailand's finance minister warns that someone used an AI face swap to impersonate him to promote an investment scam
NewsThaigerEvent
- The paper proposes APD, which distills jailbreak prompt generation to small models
NewsWeChat · Mo'anjuAttacks
- Andrew Yao, Yang Zhilin and others release the TCSExam benchmark: none of 30 open conjectures independently cracked by AI
NewsWeChat · QbitAIAI updates