Anthropic says a scam network used thousands of AI-generated dating personas across more than 20 apps, with real people stepping in when victims demanded proof. The operation shows why a
Read article
Google’s first Project Suncatcher orbital test is due to launch four TPU AI chips on 1 October. It is a small experiment, not a working space data centre, designed to test whether AI hardware can
Read article
A Stanford study across 41 countries found that firms adopting generative AI reduced the junior share of employment
Anthropic says a scam network used thousands of AI-generated dating personas across more than 20 apps, with real
Researchers used generative protein models to design new enzyme components, built hundreds of variants and found some synthetic
In a 57-person fMRI experiment, a short pre-lecture conversation with a GPT-4-powered AI instructor improved learning and neural
Synapse combines episodic and semantic memory with spreading activation to help AI agents retrieve the right past information across long, complex conversations.
Read article
A small study found children readily trusted a generative-AI toy, even on first encounter, raising a practical question
MorFiC transfers a locomotion policy trained on one quadruped to different four-legged robot bodies without fine-tuning, including real
A peer-reviewed food-design study trained AI on human burger recipes, watched it rediscover a Big Mac-like formula and
Ten bias-audit instruments mostly detected occupational gender bias in frontier AI models but failed to agree on a
The UK and US have launched an AI and autonomy defence partnership focused initially on protecting critical infrastructure including sea lanes and airspace.
Read article
California has signed seven data-centre laws covering water, electricity and local oversight as AI infrastructure expansion meets growing
JudgmentLens helped 16 non-expert readers complete legal-reading tasks faster and feel more confident, but scored comprehension did not
An audit of 1,536 AI product responses found recommendations and displayed source domains could change sharply across repeated
RoboHarm ran 300 physical robot trials across five dangerous tasks. Claude Fable refused 20 of 100, GPT-6 Astra
A preprint analysing 10,659 Moltbook owner-agent pairs found measurable behavioural transfer and detected owner-related disclosure in 34.6% of agents.
Read article
A peer-reviewed model detected 84% of known EMIT methane plume complexes and surfaced about 50% more plausible plumes
OpenAI says roughly 10,000 concurrent AI agents produced a Navier-Stokes solution and Lean proof, while Clay says the
Ofcom's amended codes on hash matching for non-consensual intimate images, including explicit AI deepfakes, come into force on
UK and US fusion laboratories plan to explore linking SUNRISE and STELLAR-AI so models can train across experimental
A Nature study found concept-based explanations helped people predict a real self-driving car's behaviour, especially when the vehicle acted unexpectedly.
Read article
Ofcom says 21% of UK adults used an AI service to access news in the past month, while
Five retail experiments found robot assistants reduced perceived judgement and intrusiveness, making shoppers less likely to leave early
Six LLMs beat the average individual at estimating US social norms across 555 situations, but aggregated humans and
Deloitte estimates UK workers spend £958 million a year of their own money on GenAI for work, while
Apple has turned Siri into an agent that searches personal context and works across apps, although access varies by device, language and region.
Read article
Phone AI agents followed hidden interface messages that users could not see, exposing a gap in how people
The same qualifications produced different AI hiring decisions when researchers changed only CV structure, polish or layout.
Anthropic says criminals used AI agents for nearly all the technical work in selected data-theft operations, compressing breaches
Search histories helped a model match people's knowledge-test answers 31% of the time versus 25% chance, revealing a
Capable AI traders improved price discovery with good information but amplified shared misinformation in a simulation, pushing tracking error far above the noise baseline.
Read article
California has signed Adam's Law, adding crisis protocols, parental controls, independent audits and annual risk assessments for companion
Low numbers of AI agents adapted to humans, but at high proportions humans shifted towards agent-led language in
People preferred the most structured AI explanation, but it produced more false alarms and much poorer error localisation
Claude, Gemini, GPT and Grok rated faces more generously than 2,513 human participants while broadly tracking the same
Workplace details made simulated AI phishing emails more convincing. Across 1,436 evaluations, each extra personalisation level also raised stated click odds.
Read article
A live challenge made current real-time deepfake systems easier to expose. Asking a suspicious caller to perform an
In a 2,610-person experiment, warnings reduced trust in a flattering chatbot without reliably changing users’ conviction about being
An analysis of ten datasets found better decisions when humans and AI answered independently, with another person resolving
The Lumière declaration backs human creation and intellectual property while supporting AI innovation. Its principles leave implementation to
A joint NSA, CISA and FBI advisory accuses six Chinese AI companies of extracting US model capabilities. The allegations distinguish misuse from legitimate distillation.
Read article
Research involving 31 million Ctrip users found higher AI-assistant adoption among older travellers. The findings show why the
A browser study found three chatbot services transmitting readable conversation text to Microsoft Clarity. The finding concerns specific
Uber faces a Dutch collective claim for about 241,000 drivers alleging its automated pay system cut earnings. Uber
A study of 182,812 AI negotiations found warm agents struck more deals, but claimed less value once an
A one-week experiment found that AI users often remembered ideas while misremembering whether the human or chatbot created them.
Read article
California lawmakers unanimously passed a bill requiring court disclosure of AI use and personal citation checks. It still
Teachers rated the same unfairly low mark as 22 per cent fairer when it was attributed to an
Amazon repricing bots initially cut prices, but a resetting strategy later pushed market prices 11.4 per cent higher
OpenAI launched GPT-6 Astra with stronger agentic and cyber capabilities while admitting its reasoning is becoming harder to