Modelos, agentes, investigación y negocio de la IA.
MIT researchers have developed DAAAM, a long-term memory system for robots that enables them to remember objects and their locations. This breakthrough could transform domestic and industrial robotics, solving a key limitation in human-robot interaction.
AWS has launched Continuum, an AI-powered security service that continuously discovers, investigates, and remediates vulnerabilities. With assisted and autonomous modes, it aims to bridge the gap between the speed of code generation and the slowness of traditional security processes.
OpenAI has introduced 'Patch the Planet', an initiative within its Daybreak program that combines artificial intelligence with human review to help open source maintainers identify and patch security vulnerabilities. The project aims to alleviate the workload of volunteer maintainers and improve the security of the open source ecosystem.
NVIDIA has presented at TM Forum's DTW Ignite 2026 event the key components for telecom operators to advance toward autonomous networks. The proposal includes synthetic data generation with NeMo Safe Synthesizer, language models specialized in the telco domain, and secure runtimes for AI agents that operate 24/7.
OpenAI unveiled Daybreak, a set of cybersecurity tools that integrate artificial intelligence to automatically find, validate, and patch vulnerabilities. It includes Codex Security, specialized in code analysis, and GPT-5.5-Cyber, a model trained on exploit data. The announcement marks a milestone in applying AI to offensive and defensive security, with implications for companies, developers, and the balance of power in cyberspace.
At Cannes Lions, NVIDIA and partners like Alembic, AWS, Criteo, and Taboola showcase technologies that bring AI to enterprise scale in advertising and marketing. From causal models to real-time bidding, the industry moves toward autonomous operations.
An artificial intelligence system has won its first trial in the UK. Garfield AI represented a freelancer in a county court, recovering £7,000 in unpaid fees and dismissing a counterclaim. The case marks a before and after in legal automation and raises questions about the future of the profession.
Anthropic updated its privacy policy to allow Claude to request age and identity verification using official documents. The measure, not yet implemented, aims to comply with regulations and mitigate risks, but raises questions about privacy and equitable access.
Google has invested about $75 million in A24, the independent studio behind hits like 'Backrooms' and 'Obsession', as part of a research alliance with DeepMind. The deal, the first in which Google acquires a stake in a film studio, aims to develop AI tools for storyboarding and pre-production, avoiding automated content generation that faces industry backlash.
Nvidia has announced a cooling system that reduces water usage in data centers. However, the largest water consumption associated with AI occurs in power plants that generate the necessary electricity, an aspect Nvidia does not address.
Anton Osika, CEO of Lovable, argues that European AI startups suffer from a confidence deficit, not a talent deficit. In a post on X, he contends that the pressure to move to San Francisco is a myth and that the real obstacle is a lack of ambition and self-confidence. This debate reopens the discussion on the European entrepreneurial ecosystem versus the American one.
Meta accidentally exposed internal data from its controversial employee tracking program, which collects keystrokes to train AI models. The incident reignites the debate on workplace surveillance and data protection.