Follow on Telegram

ai/research

Technology News

Major AI companies back a shared robots.txt opt-out signal

Google, OpenAI, Microsoft, and Anthropic are backing a proposed common signal for robots.txt that publishers could use to indicate whether their content may be scraped for AI training. The proposal, put forward by Anthropic technical lead Kevin Kelley as part of Internet Engineering Task Force work, would add a

Canada and EU form alliance spanning Ukraine, minerals and AI

Canada and the European Union have announced a new partnership combining cooperation on Ukraine with agreements involving minerals and artificial intelligence. The initiative links the bloc’s diplomatic, trade and technology agendas. Its Ukraine component may provide a framework for closer coordination, while the minerals agreements are likely intended to

SEC Sets 2027 Examination Priorities Around AI and Cyber Risk

The Securities and Exchange Commission is expected to focus on artificial intelligence and cyber risk in its 2027 examination priorities. The areas are likely to be significant for public companies navigating the increasing use of AI and growing exposure to cyber threats. The announcement signals that regulators are paying attention

Technology News

Anthropic’s IPO Filing Frames Advanced AI as Both Existential Risk and Transformative Promise

Anthropic has entered the IPO process with a filing that asks investors to consider two profoundly different outcomes for artificial intelligence: a technology that could pose “catastrophic or existential risks to humanity,” or one that could “bring forth an age of miracles.” The contrast appeared near the top of the

Anthropic Model Reportedly Identified Linux Kernel Vulnerability

Anthropic’s AI model has reportedly identified a security flaw in the Linux kernel that could enable remote root access. The finding highlights the growing use of AI systems in vulnerability research. However, the report does not specify the underlying weakness, affected Linux releases, exploitation requirements, whether the issue has

Ultracold Atoms Offer a Clue to the Emergence of Time

Researchers have used 24,000 ultracold atoms to build a miniature, isolated quantum system designed to probe one of physics’ central questions: where does time come from? The system repeatedly expands and contracts, resembling a cyclic approximation of the Big Bang followed by a possible Big Crunch. According to the

Multiscale Sampling Algorithm Improves Distance Estimates in Large Networks

A nearly 25-year-old limit in how computers estimate distances across large networks has been broken by a new algorithm developed by a computer scientist. The method provides more reliable estimates for closely connected points without sacrificing computational speed. Its key innovation is a multiscale sampling strategy that examines