Scan record
Nov 1639 loaded
- LLMs are vulnerable to malicious requests whose intent is hard to distinguish
PaperarXiv: Jailbreaks, Prompt Injection, Poisoning & DefensesThe item you came from
- Representational geometry as a diagnostic tool for adversarial robustness
PaperarXiv: 其他 AI 安全论文
- Forescout researchers used AI to discover security vulnerabilities in discontinued devices that no one is patching
NewsHelp Net Security AIAttacks
- Western Cape Department of Infrastructure (DOI) and AIFOD explore responsible AI in infrastructure
NewsWCGAI updates
- Amazon says its honeypot captured records of AI attack agents at work, with attackers using stolen API keys to drive attacks and search for more keys
NewsTechnoEdgeEvent
- The Japanese government is considering establishing new penalties in the revision of the AI law for companies developing high-performance AI, to punish violations of orders
NewsinnovaTopiaGovernance
- Top scorer in South Korea's tax accountant exam hands exam-related devices to police in AI cheating case
NewsSeoul Economic DailyEvent
- Korean education YouTuber and lecturer call for conditional postponement of CSAT
NewsChosunBizEvent
- Australia plans to establish a frontier AI safety regulator and mandate incident reporting
NewsCSTIGovernance
- The top scorer in South Korea's certified tax accountant national examination has confessed in writing to police, detailing how he cheated using a miniature camera and AI
NewsKorea JoongAng DailyEvent
- Kim Min-seok, leader of South Korea's Democratic Party: studying extending exam ban for AI cheaters from 5 years to 10 years
NewsFinancial News KoreaGovernance
- Four days after Jensen Huang stepped down, talks of his return are reported, with over 150,000 people signing a petition supporting his return
NewsSBSAI updates
- AI agent prompt injection and key exfiltration: Stroq open-source test, 36 attack scenarios, 0 penetrations
NewsWeChat · Goumang Security LabTools
- Cursor affected by the Fireworks AI intrusion incident; a malicious version of the Claude Code extension was once listed on its marketplace
NewsiThome (Taiwan)Event
- Uber: personal ID information of thousands of drivers leaked due to the Fireworks AI hack
NewsiThome (Taiwan)Event
- The EU has always found a way out, but on AI it struggles to see a direction
NewsPOLITICOGovernance
- Car owner says Claude Code used a $3 Arduino to crack and reverse-engineer the engine computer of a 32-year-old Jeep
NewsCybersecuritynewsAI updates
- South Korea's National Assembly Education Committee to hear a report tomorrow on countermeasures against AI glasses cheating, focusing on anti-cheating measures for the College Scholastic Ability Test
NewsSBSGovernance
- Reform Party's Lee Ki-in says postponing the college entrance exam should be considered in the worst case
NewsFinancial News KoreaGovernance
- Calls in the tech world for Jensen Huang to return to Nvidia are growing louder
NewsCNBC · TechnologyAI updates
- PolicySniff preprint: agent approval policies leak on shared GPUs
NewsCASRAIAttacks
- Study finds RAG citations often reflect model parametric memory rather than retrieval sources
NewsCASRAIEvaluation
- Hannah Li's study: why user adaptation distorts LLM benchmarks and A/B tests
NewsCASRAIAlignment
- OpenAI discloses Control Room: can stop research workloads from continuing to obtain model outputs
NewsCASRAIDefenses
- Mirae Asset Chairman Park Hyeon-joo co-signs petition demanding Nvidia's board restore Jensen Huang's CEO position before the independent investigation is completed
NewsSeoul Economic DailyAI updates
- Detection equipment to prevent AI exam cheating will not be ready for South Korea's CSAT
NewsKorea JoongAng DailyEvent
- Small, cheap and hard to detect, AI cheating tools challenge South Korean exams
NewsKorea JoongAng DailyEvent
- Musk joins an open letter with more than 165,000 signatories calling on Nvidia to reinstate Jensen Huang
NewsThe Guardian · AIAI updates
- South Korean Prime Minister vows to mobilize national forces to prevent AI cheating on the college entrance exam
NewsYonhapEvent
- The author criticizes the AI safety community: on one hand calling for more people to join, on the other shutting outsiders out
NewsLessWrongOpinion
- ARTEX-IOT open-sourced: a multi-agent autonomous discovery system for IoT firmware auditing
NewsWeChat · SecureNexusLabTools
- Aikido says hackers used 4,837 compromised GitHub accounts to probe the write permissions of 60,000 repositories
NewsiThome (Taiwan)Event
- FBI warns that hackers are using the AI tool ARTEX to accelerate intrusions, with some attacks gaining high privileges in under 3 hours
NewsiThome (Taiwan)Event
- Evoke Security isolated experiment: a code change lets the caller decide the evaluation rules
NewsWeChat · Oxo SecurityAttacks
- Geoff Stearns predicts: agents will prefer devices with flashable firmware
NewsGeoff StearnsOpinion
- First place in South Korea's 63rd tax accountant exam admits to cheating with generative AI, has surrendered to police
Newsnewsworks.co.krEvent
- Data shows AI-related child sexual abuse material reports grew more than 8,400% in two years
NewsThe Guardian · AIEvent
- Institute of International Socio-Economic Research report: two major topics, frontier model access safeguards and loss-of-control risk
Newsi-ise.comGovernance
- The key to AI agent identity is who performed the operation, not who logged in
NewsIslandOpinion