AI

Microsoft’s CEO wants advanced models treated like potential insider threats: controls outside the model, tamper-proof logs, and a switch to stop a task mid-run.

The Wikimedia Foundation says it found OpenAI agents active on its platforms: millions of requests, odd edits to a citation tool, and a possible link to May’s Wikidata disruption.

New computer vision systems can track employee desk time, generating detailed reports. But this technology raises legal and ethical questions: what do European regulations say? And are we sure that "presence" equals "work"?

At Trump's urging, the US administration has replaced "Artificial Intelligence" with "Super Intelligence" (OSI). This move goes beyond semantics, reshaping public perception and placing AI at the heart of geopolitical competition.

Agentic AI doesn’t promise “same command, same result.” It’s probabilistic and context-bound. Real workflows need governance: roles, guardrails, human review. Orchestration, not autopilot, is how digital work will evolve.

As the tech race heats up, the US and China are initiating a dialogue on AI safety. The goal is to create a direct channel to prevent catastrophic incidents, acknowledging shared risks that transcend their fierce competition.

Singapore switched on a prototype “biocomputer” using lab-grown human neurons connected to chips. Researchers probe efficiency, ethics, and data center impact. If a computer is alive, it must be kept alive: temperature, oxygen, nutrients.

Anthropic’s Dario Amodei urges the AI sector to slow down and adopt independent oversight, shared standards, and international accords. With Sam Altman and Elon Musk signaling caution, the next AI wave demands shared responsibility.

Since 04/09/2026, dozens of Tesla “Cybercab” vehicles without steering wheels or pedals operate in Austin. Tesla bets on camera-only autonomy and neural nets. Public rides reignite debates on safety, regulation, and AI’s impact on work and mobility.

Studies show AI-assisted texts are rated higher on quality and creativity, yet they cluster around similar phrasing and ideas. The average improves, the extremes fade. What it means for digital ecosystems and research.

SpaceX e Nvidia annunciano il primo rack di calcolo orbitale entro fine 2027. Energia solare continua, ma raffreddamento per irraggiamento, radiazioni e manutenzione remota restano nodi. Tra promesse e limiti fisici, cosa cambia per il digitale e l’AI?

By 2030, the United Nations estimates, data centres could reach 945 terawatt-hours a year. And daily use, not model training, is the real culprit.

The future is already here. Generative AI is set to radically transform work and society. As technology advances at a dizzying pace, global institutions like the EU with the AI Act are trying to define the rules for ethical and secure innovation.

The advance of AI promises to democratize once-elite skills, from art to programming. But, like past industrial revolutions, it could eliminate specific professions. It is unstoppable progress, but who will pay the price?

As of August 2, 2026, the EU AI Act requires invisible, machine-readable marks in AI-generated content. Anthropic ships them in Claude; Google applies SynthID in Gemini. The signature travels with words but weakens on short or heavily edited texts. Impacts: traceability, accountability, misattribution risks.

As governments and tech giants like Clearview expand biometric surveillance with databases of billions of images, innovation is fighting back. From Italy, startups like Capable are creating textiles that deceive AI, protecting our identity and privacy.

A U.S. startup is testing a wave-powered, ocean-cooled floating data center. If it withstands storms, corrosion and costs, it could ease grid bottlenecks and reshape AI scaling from 2027, with off-grid compute beamed back by satellite.

A super-AI, Mythos, has entered US-China negotiations for the first time. While Beijing fears its use as a cyber weapon, Washington treats it as a strategic technology, restricting its access just as it did with chips.

Anthropic shifts MCP to a stateless HTTPE model with asynchronous tasks and in-chat UI apps, promising simpler scalability, lower infrastructure costs, and secure instant integration for users, enterprises, and developers.