CIRCLE WITH A DOT

Oh, so now your mouse curser is being used to train the dumb AI so that it can understand the curser you've used correctly for decades without an AI middleman. "Essentially, rather than relying on an AI model to consistently tell the difference between a shopping list, a recipe, a food fight, and a hamburger costume, a combination of cursor gestures and naturalistic commands like 'move this here' are leveraged to point the AI in the right direction."PC Gamer: Google DeepMind wants to reinvent the humble mouse cursor https://www.pcgamer.com/software/ai/what-if-behind-the-pointer-there-was-an-ai-model-google-deepmind-wants-to-reinvent-the-humble-mouse-cursor/ #Google #DeepMind

RT @SciTechera: Wissenschaftler sagen, der nächste KI-Durchbruch sind nicht größere Modelle, sondern intelligentere. Neuro-symbolische KI: In einer neuen Analyse betonen Forscher den Aufstieg der neuro-symbolischen KI, eines hybriden Ansatzes, der neuronale Netze mit formaler Logik und regelbasierten Systemen kombiniert. Anstatt sich rein auf Wahrscheinlichkeiten wie LLMs zu verlassen, nutzen diese Systeme strukturierte Reasoning-Pipelines, bei denen Ergebnisse validiert, eingeschränkt und logisch konsistent sind. Aktuelle Systeme von Google DeepMind zeigen, wie das funktioniert: AlphaGeometry 2 (2-2026) löst ca. 83–88% der Geometrieaufgaben der International Mathematical Olympiad, einige Lösungen in Sekunden AlphaProof (2024 → aktuell) erreichte 28/42 Punkte auf IMO-Niveau durch formal verifizierte Beweise AlphaFold vorhergesagte über 200 Mio. Proteinstrukturen mit nahezu experimenteller Genauigkeit, was zeigt, wie hybride KI wissenschaftliche Probleme skalier a lösen kann. Dieser subtile, aber massive Wandel deutet darauf hin, dass KI von rein probabilistischem Textgenerieren zu echtem Reasoning mit verifizierbarer Korrektheit übergeht – ein Fortschritt, der Wissenschaft, Mathematik und Ingenieurwesen neu definieren könnte. mehr auf Arint.info #ArtificialIntelligence #DeepMind #KI #MachineLearning #TechTrends #arint_info https://x.com/SciTechera/status/2046294388247245299#m

RT @HowToAI_: Google DeepMind just dropped the most terrifying cybersecurity paper of the year. They just mapped the attack surface that nobody in AI is talking about. Websites can already detect when an AI agent visits and serve it completely different content than humans see. - Hidden instructions in HTML. - Malicious commands in image pixels. - Jailbreaks embedded in PDFs. This “detection asymmetry” means a site can serve normal content to you, and malicious, hidden content to your agent. The agent doesn’t know it’s being tricked. It simply processes whatever it receives and acts on it. Here’s the attack surface nobody is talking about: → Indirect Web Injection: Malicious instructions hidden in HTML comments, CSS tricks, or white text on white backgrounds. → Multimodal Steganography: Commands encoded directly into image pixels, invisible to humans, but fully readable by vision models. → Document Jailbreaks: Override instructions embedded deep inside PDFs, spreadsheets, and calendar invites. → Memory Poisoning: Injecting false information that persists across future sessions. → Exfiltration Attacks: Tricking the agent into sending your private data to attacker-controlled endpoints. → Multi-Agent Cascades: The worst-case scenario, Agent A gets compromised, passes the “poison” to Agent B, then to Agent C. The entire pipeline gets infected because agents trust each other’s data. The most sobering part of the DeepMind report? The defense landscape is failing, badly. Input sanitization doesn’t work because you can’t “sanitize” a pixel. Prompt-level instructions to “ignore su… Arint.info #Agent #agent #AIagent #DeepMind #Google #scale #arint_info https://x.com/HowToAI_/status/2045749883773333717#m

Been leaning on Gemma 4 for most of my conversational tasks and some light tool calling this week. Honestly? Surprisingly competent.#Gemma4 #DeepMind #agenticAI #LLMs #ai