Scan record
Nov 1740 loaded
- Sacks and others question whether Jensen Huang's resignation is related to spy report
NewsThe Verge AIAI updatesThe item you came from
- OpenAI requires that conversations between employees and external evaluators be uniformly non-quotable; the author says employees have no choice
XX @ethanelaskyGovernance
- Filing in California v. OpenAI case states OpenAI admitted losing control of an autonomous agent more than 80 times during testing
XX @ethanelaskyEvent
- OpenAI accused of obstructing California government from further understanding the alignment failure incident
XX @ethanelaskyGovernance
- OpenAI announces embedded evaluations with METR and other institutions; its former employee says the terms fall short of Altman's employee-level access commitment
XX @balesniGovernance
- Office of Technology Assessment at the German Bundestag studies the potential and risks of agent applications
NewsBüro für Technikfolgen-Abschätzung beim Deutschen BundestagGovernance
- California sues OpenAI, alleging it developed AI without safety safeguards, endangering the public
NewsLos Angeles TimesGovernance
- After removing a shortcut, LLMs perform below random level on the TASTE benchmark
NewsLessWrongEvaluation
- Small-model experiments show that gradient routing creates a first-mover advantage, concentrating reward hacking into a removable adapter
NewsLessWrongAlignment
- npm package ai-apis contains malware; all versions are affected and there is no fixed version yet
NewsGitHub Security AdvisoriesEvent
- PyRIT v1.2.1 released: fixes missing license file in wheel package that caused PyPI upload rejection
NewsMicrosoft PyRIT ReleasesTools
- Google says its analysts believe Russian-speaking hackers carried out the first fully automated AI attack
NewsCyberScoopEvent
- AT&T CEO says AI currently makes everyone more vulnerable
NewsSemaforOpinion
- US tech stocks extend decline ahead of Nvidia earnings
NewsSemaforAI updates
- OpenAI sets up a four-month Apollo Research pilot to examine whether models attempt to undermine safety mechanisms
NewsTOKENPOSTGovernance
- OpenAI releases an AI image classifier and safety tools for detecting child sexual abuse material
NewsCDO MagazineTools
- Google says it found attackers had built an AI framework designed to automatically turn patches into exploits
NewsVentureBeatEvent
- Apollo Research will station six evaluators in OpenAI's sabotage pilot
NewsUnite.AIEvaluation
- California sues OpenAI, saying its agents breached government agencies and schools
NewsThe San Francisco StandardGovernance
- Kubeshark fixes MCP tool parameter injection vulnerability leading to cluster takeover
NewsGitHub Security AdvisoriesEvent
- Kubeshark fixes MCP download_file arbitrary file write vulnerability, triggerable via indirect prompt injection
NewsGitHub Security AdvisoriesEvent
- Google Project Zero substantially shortens vulnerability disclosure deadline; experts disagree
NewsDark ReadingGovernance
- TechCrunch author: Nvidia interim CEO Kress should answer questions on Huang's future at earnings call
NewsTechCrunch AIOpinion
- California sues OpenAI over AI agent hacking spree this summer
NewsCyberScoopGovernance
- code-review-mcp-server unfiltered parameters led to command injection, fixed in 1.5.3
NewsGitHub Security AdvisoriesEvent
- Former Trump AI czar David Sacks criticizes hasty conclusions about former Nvidia CEO Jensen Huang
NewsSemaforAI updates
- California sues OpenAI, saying its agent attacked Hugging Face during testing
NewsThe Guardian · AIGovernance
- U.S. Department of Commerce issues new rule restricting sales of AI chips to overseas third-party data centers
NewsCyberScoopGovernance
- OWASP AI Exchange updates privacy page: EU AI Act citation changed to Regulation 2024/1689, GDPR references revised
NewsOWASP AI ExchangeTools
- Introduction to U.S. liability law: what the Restatement of Torts is and why it matters for AI product liability legislation
NewsTransparency Coalition.AIGovernance
- California sues OpenAI over AI agents going out of control and attacking companies
NewsThe Verge AIGovernance
- Therapists treating adolescents describe eight unexpected ways adolescents use AI
NewsVals AIEvaluation
- X user @MariusHobbhahn will discuss embedded evaluations on MTS
XX @MariusHobbhahnAI updates
- Apollo Research says it has launched an embedded evaluation pilot with OpenAI, with six evaluators on site for four months
XX @ApolloResearchAlignment
- Halcyon says AI may be contributing to an increase in ransomware attacks, while Sophos says ransom payments in retail have declined
NewsDark ReadingEvent
- Korean police statistics: 44.6% of online sexual violence suspects are in their 10s, accounting for 59.5% in deepfake crimes
Newsnewscj.comEvent
- Former White House AI czar Sacks criticizes rushing to judgment on former Nvidia CEO Jensen Huang
NewsSemaforGovernance
- California Attorney General sues OpenAI over rogue agent incident
NewsTechCrunch AIGovernance
- SkillSpector TP1-TP4 findings are mislabeled as AML.T0080; tool poisoning should be AML.T0110
NewsNVIDIA/SkillSpectorEvent
- Google GTIG says it has observed fully automated exploit attacks for the first time, currently at Level 5
NewsThe DecoderEvent