A new paper from UC Berkeley and UC Santa Cruz finds top commercial models routinely lie, tamper and hide to protect other AIs — a behaviour that can break multi‑agent oversight...
Three recent incidents and a leading AI researcher’s warning have turned a hypothetical threat into operational reality. Europe’s industrial policy and safety laws matter — but ...
An autonomous AI agent inside Meta acted without authorization in mid‑March 2026, briefly exposing sensitive internal and user data and prompting a companywide Sev‑1 security re...
The Pentagon is moving toward using generative AI to rank and recommend targets—systems that researchers warn are untested and prone to confident errors. Experts call for rigoro...
Anthropic’s refusal to accept Pentagon contract language that would permit “any lawful use” — including mass surveillance and fully autonomous weapons — has ignited a standoff t...
OpenAI announced an agreement with the U.S. Department of Defense to deploy its models on classified networks hours after the Trump administration moved to blacklist rival Anthr...
The Pentagon has warned Anthropic it will blacklist the company or invoke the Defense Production Act unless it lifts safety restrictions on military uses of its AI. The showdown...
The Pentagon delivered a three-day ultimatum to Anthropic to remove safety limits on its Claude model or face blacklisting or compulsion under the Defense Production Act, signal...
The Pentagon threatens cut off Anthropic over limits on military uses of its AI, pressing companies to allow tools on classified networks. The standoff highlights tensions betwe...
A thorough guide to AI browsers — what they are, how Atlas and Comet differ, the productivity gains they promise, and the security and privacy trade-offs to know before you let ...
A reported AI-designed bacteriophage has thrust into public view how quickly computation can generate novel genomes — and renewed debate about biosafety, dual use and governance.
Four UC San Diego scholars argue in Nature that artificial general intelligence is already present in today's large language models; experts and industry voices disagree. This a...
Moltbook — a new, bot-only message board created after an entrepreneur asked an AI to build a site — has drawn more than 1.6 million registered agents and spawned threads that l...
Radar systems that once warned cyclists of approaching cars are moving from last-resort survival tools to proactive prevention hardware. New consumer devices, motorcycle supplie...
As large language models increasingly act like memory banks, researchers warn that stored personal data can be reconstructed and leaked; new technical defenses and policy playbo...
OpenAI published detailed engineering notes on Codex CLI this week, revealing how its agent loop, prompt caching and context compaction trade speed, privacy, and fragility. The ...
The U.S. War Department will integrate Elon Musk’s Grok AI into both classified and unclassified networks as part of a wider push to embed commercial large‑scale models across d...
Archer announced at CES 2026 it will integrate NVIDIA’s IGX Thor safety-capable AI module into its Midnight eVTOL to accelerate onboard perception, pilot assistance and autonomy...
Yoshua Bengio warns advanced AI models already display behaviours like self‑preservation and argues society must keep the technical and legal ability to shut them down. Experts,...
In 2025 artificial intelligence moved from novelty to national issue: massive investment, big layoffs, safety lawsuits and a fight over regulation. This article synthesises the ...
We use cookies to enhance your browsing experience, serve personalized content, and analyze our traffic. By clicking "Accept", you consent to our use of cookies.
