intelligent systems

Bill Gates has described artificial intelligence as the most dangerous technology ever invented, a striking shift for one of AI's most prominent advocates. His comments come amid intensifying global debate over AI safety, regulation and the risks advanced systems could pose to jobs, security and society.

The US Federal Trade Commission has opened an investigation into the safety of artificial intelligence, covering leading AI developers including OpenAI and Anthropic. The probe signals growing regulatory scrutiny of how major AI companies build and deploy their systems. News of the inquiry is drawing attention as regulators weigh potential consumer protection and safety concerns in the fast-moving AI sector.

The chief executive of Anthropic has called for stronger government regulation of artificial intelligence, in an exclusive interview with ABC News. The company behind the Claude chatbot argues that clear rules are needed to manage the risks of increasingly powerful AI systems. The remarks add Anthropic's voice to an industry-wide debate over how far regulators should go.

Microsoft's president has warned that artificial intelligence faces weaker oversight than everyday products, saying AI is less regulated than milk, and is calling for tougher safety rules. The comparison is drawing attention as governments worldwide debate how to regulate rapidly advancing AI systems without stifling innovation.

The United States has banned Chinese AI firm DeepSeek from government agency systems, according to CNBC. The move comes amid rising tensions between Washington and Beijing over artificial intelligence, and marks a significant escalation in the tech rift between the two countries. The ban restricts the popular AI model from being used on official US government devices and networks.

The U.S. Department of Homeland Security has issued an order banning Chinese AI firm DeepSeek from use across federal civilian agency IT systems. The move follows mounting U.S. concerns over security and data risks tied to Chinese-developed artificial intelligence tools, and signals a tightening stance on foreign AI software in government use.

Physical Intelligence has opened a beta of its API, giving developers access to its frontier robotics models for controlling robots. The company, which builds general-purpose models for physical tasks, says the API will let teams apply the same models across different robots rather than training bespoke systems. Commenters are debating whether generalist robot models can match task-specific controllers and how the offering compares with rival robotics AI platforms.

Chinese AI lab DeepSeek has released a new open-source tool designed to reduce the operating costs of running artificial intelligence systems. The move, reported by Bloomberg, continues DeepSeek's strategy of sharing efficiency-focused technology openly, challenging rivals that keep such tooling proprietary and underlining its focus on making AI inference cheaper.

A piece attributed to Rich Hickey, the software designer best known for creating the Clojure programming language, makes the argument that artificial intelligence should not be described as a tool. The essay is drawing attention among developers and technologists, who are debating whether that framing changes how AI systems should be built, evaluated and trusted. Hickey's views on software design have long carried weight in programming communities.

Digital rights organisation Open Rights Group is launching a report titled 'Automated Discrimination: Unpacking the Digital Hostile Environment' on Wednesday 28 October. The report examines how artificial intelligence is used within the UK immigration system, the secrecy surrounding those tools, and sets out recommendations for reform. The launch comes amid growing scrutiny of algorithmic decision-making in government, particularly where automated systems affect vulnerable people's rights and access to justice.

Quanta Magazine reports that an AI system has produced a proof of the quasi-Riemann hypothesis, a mathematics result related to the famous Riemann hypothesis about the distribution of prime numbers. The article asks what this milestone means for the future of machine-assisted mathematical research, and mathematicians and technologists are debating whether AI is now capable of genuine theorem-proving rather than just assisting human mathematicians.

A new discussion of Texas's TRAIGA law is examining who bears legal responsibility when artificial intelligence systems make mistakes. The session frames 2026 as a pivotal year for AI policy, with the Texas act serving as a case study in how US states are drafting liability rules for automated decisions and errors. Legal observers and founders are weighing what compliance will require as more states move to regulate AI.

Mexico's Financial Intelligence Unit (UIF) says the FATF (GAFI) evaluation of the country's money laundering prevention system is in its final stage. The unit has been central to a reported financial crackdown on the Sinaloa Cartel's network, with US Ambassador Ronald Johnson describing it as a hard blow to the cartel's criminal and political financing. Meanwhile, President Sheinbaum clarified that the UIF detected anomalies but no criminal links in a case drawing scrutiny.

Japan's Leading-edge Semiconductor Technology Center and partner organisations have announced plans to jointly develop an artificial intelligence-based system for processing medical images. The collaboration aims to bring semiconductor expertise into healthcare, using AI to analyse scans and support diagnosis. The announcement has drawn attention for linking Japan's chip industry with medical technology.

Defence officials under Mexico's Security Cabinet, led by figure Ricardo Trevilla Trejo, announced that the National Guard will be connected to a centralised criminal intelligence system. The measure is part of the government's push to coordinate security forces and improve information-sharing in the fight against organised crime. Local media are reporting the integration plan and what it means for public security operations.

The US Navy has expanded its robot ship program, with the deal quadrupling in size as part of an effort to 'outsmart' China using artificial intelligence at sea. The move signals Washington's growing investment in unmanned vessels and AI-driven naval capabilities amid intensifying military competition with Beijing in the Pacific.

The Wall Street Journal reports that artificial intelligence has been used to discover a software bug that could put as many as 70% of the world's websites at risk. Details of the flaw, who found it, and which systems are affected were not included in the available reporting. The story is drawing attention for its suggestion that AI tools are now capable of surfacing vulnerabilities at internet scale.

The New York Times reports that artificial intelligence has discovered a critical security flaw affecting the technology underpinning websites. The finding highlights how AI systems are increasingly being used to identify vulnerabilities in core internet infrastructure. Details of the flaw, who exploited or disclosed it, and its real-world impact have not yet been widely reported, but security experts are expected to scrutinize the disclosure closely.

OpenAI says its artificial intelligence systems uncovered one of the most significant security vulnerabilities ever found on the web. The claim, reported by Forbes, suggests AI tools can now play a serious role in discovering critical flaws at internet scale. Details on the specific weakness, who was affected, and how it was patched have not been made clear in the reporting available.

Researchers report that artificial intelligence tools helped identify a critical security flaw affecting large parts of the internet's infrastructure. The finding highlights a growing role for AI in vulnerability discovery, as defenders rush to patch affected systems before exploitation. Security experts are weighing both the promise of AI-assisted research and the risks if similar flaws are found by malicious actors first.

OpenWorld has signed a term sheet for a financing program of up to $1 billion to build a distributed, solar-powered AI computing network. The plan would initially integrate the computing nodes with residential battery storage systems, with the financing structured as a tokenized program. The announcement combines themes of renewable energy, artificial intelligence infrastructure and crypto-based fundraising.

A new report highlights the development of artificial intelligence systems that can grasp physical laws rather than just pattern-match from data. Such models could improve engineering design, robotics and scientific simulation by predicting how objects behave in the real world. Coverage of the topic is circulating among technology and engineering audiences this week.