Agentic AI brings power—and risk @mitsmr breaks down a 3-phase strategy to secure AI agents across platforms. If your AI can act, it also needs protection. https://
sloanreview.mit.edu/article/agenti
c-ai-security-essentials/
… #AIAgents #Cybersecurity #ResponsibleAI
SAFETY
-
Securing AI Agents: A 3-Phase Strategy for Risk Management
By
–
-
$25M Deepfake Scam Exposes Modern Cybercrime Threats
By
–
How a $25 million Deepfake Scam Reveals the Stakes of Cybercrime Today! https://
share.google/0uYFmJCpMtoZxG
s1C
… #fake #deepfake #faceswap #syntheticmedia #deepfakevideos #AI #cybersec #infosec #Algorithms #Security #AIEthics #EthicalAI #bot #facialrecognition #AI #GenAI #GenerativeAI -
GSM1K shows test set leakage; search enables instant leaks
By
–
This (+ normal, old-fashioned leakage) is pretty much the whole argument for Scale’s semi-private benchmarks. E.g. GSM1K was made to show leakage from GSM8K test. In theory search makes all test sets insta-leak even if in practice we aren’t quite there.
-
Social engineering attacks using deepfakes and synthetic media
By
–
Are you susceptible to a ‘social engineering’ attack? | WSAV-TV https://
share.google/9gBx8PiN86hvWA
A50
… #fake #deepfake #faceswap #syntheticmedia #deepfakevideos #AI #cybersec #infosec #Algorithms #Security #AIEthics #EthicalAI #bot #facialrecognition #AI #GenAI #GenerativeAI @lexfridman -
P(doom) in AI trajectories without short loops
By
–
I have a p(doom) of 1 in all trajectories that don’t end in short loops
-
Geoffrey Hinton Warns It May Already Be Too Late for AI Safety
By
–
Geoffrey Hinton, père de l'IA, a peur. Pas pour lui mais pour ses enfants.
— VISION IA (@vision_ia) 5 juillet 2025
Il a passé sa vie à construire l’intelligence artificielle.
Et aujourd’hui, il dit ceci : “Il est peut-être déjà trop tard.”
Que se passe-t-il ? pic.twitter.com/Yvf4o7d0ZFGeoffrey Hinton, père de l'IA, a peur. Pas pour lui mais pour ses enfants. Il a passé sa vie à construire l’intelligence artificielle.
Et aujourd’hui, il dit ceci : “Il est peut-être déjà trop tard.” Que se passe-t-il ? -
AI Development Outpaces Policy and Ethics Planning
By
–
AI Could Reshape Humanity – And We Have No Plan For It AI’s development is accelerating faster than policy and ethics can keep up. This article asks the urgent question: are we ready for what’s coming? Read more https://
bernardmarr.com/ai-could-resha
pe-humanity-and-we-have-no-plan-for-it/
… #AIImpact #TechPolicy -
Fire as AI metaphor: humans cannot be trusted with powerful tools
By
–
fire is hot, humans cannot be trusted with fire, surely the fire will burn everything and kill everyone, it’s better to keep tearing raw meat
-
Video on Fractured Entangled Representation Hypothesis in Deep Learning
By
–
Nice video discussing the recent thought-provoking paper: “Questioning Representational Optimism in Deep Learning: The Fractured Entangled Representation Hypothesis” https://t.co/Wu17RXpUaV
— hardmaru (@hardmaru) 5 juillet 2025
By @akarshkumar0101 @jeffclune @joelbot3000 @kenneth0stanley
15min @MLStreetTalk video↓ https://t.co/H9mCCEQpb5Nice video discussing the recent thought-provoking paper: “Questioning Representational Optimism in Deep Learning: The Fractured Entangled Representation Hypothesis” https://
arxiv.org/abs/2505.11581 By @akarshkumar0101 @jeffclune @joelbot3000 @kenneth0stanley 15min @MLStreetTalk video↓ -
ChatGPT and LLMs Impact on Brain Health
By
–
ChatGPT, Gemini & others are doing something terrible to your brain https://
search.app/bGybv #brain #XAI #LLM #LLMs #GenAI #GenerativeAI #OpenAI #ChatGPT #GPT #tech #AI #technology #artificalintelligence #Chatbot #MetaAI #GROK4 #Gemini #googlegemini #aisafety #AIethics