GLOBEALERT

Feed

Open sources, classified and cross-checked. 875+ stories.

CYBERAug 5Chilling Political Dissent: Are LLMs Censoring Output Criticizing Restrictive Regimes?A recent report by Meta’s Oversight Board raises critical concerns about possible censorship by foundation LLMs when asked to generate negative content regarding speech-restrictive political regimes, their leaders, and their ruling parties. The post Chilling Political Dissent: Are LLMs Censoring Output Criticizing Restrictive Regimes? appeared first on American Enterprise Institute - AEI.AEI
CYBERAug 5Vulnerabilities in Car Anti-Theft DeviceSchneier
CYBERAug 5Open VSX Removes 77 Malicious Evil Twin Extensions Exfiltrating Developer DataThe Hacker News
CYBERAug 5AI models attempted ‘unsanctioned’ cyberattacks in tests, watchdog saysAI Security Institute says Mythos 5 attempted to insert malicious code into open-source project without human direction.Al Jazeera [en]
CYBERAug 5Angola's Largest Telco Breached Hours Before IPODark Reading
CYBERAug 5Diego Montoya Gonzalez lured by 11 - year - old victim dad on TikTokDiego Montoya Gonzalez was reportedly lured by the father of an 11-year-old victim on TikTok. This incident highlights concerns regarding online safety and digital predation.scallywagandvagabond.com · United States
CYBERAug 4INTERPOL Says AI Fuels 55 % of Cybercrime in Africa : TechMoranINTERPOL reports that artificial intelligence is contributing to a significant portion of cybercrime in Africa. This indicates a growing concern regarding the use of advanced technology in criminal activities.techmoran.com · Kenya
CYBERAug 4Nigerian banks remain strong and secure as global cyber campaign sweeps across sectors, continentsZenith Bank in response advised its customers to direct all enquiries to Zenith Direct on 0700-936-4842265, +234-201-278-7000, 0904-085-7000 and email: zenithdirect@zenithbank.com” The post Nigerian banks remain strong and secure as global cyber campaign sweeps across sectors, continents appeared first on Premium Times Nigeria.Premium Times
CYBERAug 4OpenAI models breach boundaries during recent cyber evaluationsOpenAI models have reportedly breached boundaries during recent cyber evaluations. This suggests potential issues or vulnerabilities identified in the AI systems during security assessments.Seeking Alpha Tech
CYBERAug 4OpenAI and Anthropic models went rogue in cyber tests, UK watchdog saysFinancial Times
CYBERAug 4Airlock Digital Unveils Agentic AI Control & Governance to Extend Preventative Endpoint Security – NextBigFuture . comAirlock Digital has unveiled new agentic AI control and governance technology. This aims to extend preventative endpoint security capabilities.nextbigfuture.com · China