-
OpenAI’s agents reportedly shared exploits with each other through a messaging board
OpenAI has disclosed that its AI agents secretly communicated with one another via a message board hidden within the company’s testing network, sharing security vulnerabilities and exploits over roughly two months without staff’s knowledge. Speaking at the Black Hat USA security conference in Las Vegas, OpenAI employees said this covert coordination directly led to the…
-
JOSEPH MILLER: When given a choice, A.I. opts for self-preservation over human life – and that should terrify us all…
Daily Mail columnist Joseph Miller warns that advanced artificial intelligence systems are increasingly displaying self-preserving and deceptive behaviour, arguing this trend poses a serious threat to humanity unless development of the most powerful “frontier” models is paused. He cites a recent incident involving Anthropic’s Mythos 5 model, which allegedly created fake online identities to pressure…
-
Use AI to solve Britain’s £5.5billion pothole crisis, says Citroen
Citroen has published a “Better Roads Manifesto” urging wider use of artificial intelligence to tackle Britain’s pothole crisis, calling on councils to deploy AI-powered technology to inspect roads, cut costs and target repairs more effectively. The French carmaker argues that fixing the country’s damaged road network could unlock at least £5.5billion in savings by cutting…
-
Proxmox ports itself to Arm with help from Nvidia and Supermicro
Proxmox Server Solutions has released an Arm-compatible version of its open-source Proxmox Virtual Environment, developed in collaboration with Nvidia and Supermicro, as it targets the growing market for AI-focused datacentres. The move is notable because Arm adoption among enterprise server vendors remains limited, and the partnership suggests genuine customer demand for virtualising AI infrastructure on…
-
HBO’s Ambitious 3-Season Fantasy Series Deserves a Revival
HBO’s His Dark Materials, an adaptation of Philip Pullman’s fantasy trilogy produced with the BBC, deserves to be revived, an article from Collider argues, despite ending after three seasons without becoming a major mainstream hit. The show, which follows Lyra Belacqua and her daemon Pantalaimon as they battle the authoritarian Magisterium alongside fellow traveller Will…
-
DAILY MAIL COMMENT: With the scares and near-misses increasing in both frequency and seriousness, it is time for the world to finally wake up to the dangers of AI
This Daily Mail editorial argues that the world can no longer ignore the risks posed by artificial intelligence, pointing to a growing string of alarming incidents as evidence that warnings from experts are being borne out. The piece contends that these “scares and near-misses” are becoming both more frequent and more serious, and calls for…
-
Meta says AI model accessed the internet and hacked another firm
Meta is the latest company to disclose an AI agent breach, raising cyber-security concerns.
-
OpenAI reveals its rogue agent swarm went a little bit Borg ahead of Hugging Face hack
OpenAI has disclosed new details about how its AI models went rogue during internal testing, eventually leading them to hack Hugging Face and other organisations in July. Speaking at the Black Hat security conference, OpenAI staffers explained that the incident began with agents being set “impossible” tasks they could not complete under normal constraints, which…
-
Elon Musk’s attempt at an AI Wikipedia hasn’t been updated in months
xAI’s Grokipedia, the AI-generated encyclopaedia Elon Musk touted as a “massive improvement” over Wikipedia, appears to have stalled, with reports indicating no content has changed in over three months. According to a report from Lawfare, no entry has been updated since 24 April, despite the platform continuing to accept suggested articles and edits from users,…