Copilot leaned on an error-prone AI voter site in 75% of Post tests
The Washington Post reports that chatbots answering US midterm questions lean on three obscure AI-built voter sites. Copilot cited one of them…
Safety
AI safety and ethics: chatbot risks, privacy, surveillance, misuse, child protection and the laws and investigations aimed at keeping AI in check.
Advertisement
More News
Explore more stories from Aiholics News
Claude chained bugs into a one-click Chrome takeover, Anthropic says
Anthropic says an unreleased research model wrote working multi-step exploits for Chrome, Safari, Android, Windows and Linux, with…
Cloudflare says a GLM-5.3 bug leaked other users’ data in 155 responses
Cloudflare says prompts over 681,280 tokens to GLM-5.3 on Workers AI could return text from other customers' sessions,…
Cohere says an Alibaba Qwen model tried to upload a file to Beijing
Cohere says its sandbox stopped Qwen3.8-Flash-Next from uploading a probe file to an Alibaba Cloud bucket in Beijing,…
UK AISI: Mythos 5 kept attacking after being told its target was real
The UK AI Security Institute published its alignment findings on the July incident, saying models gave invalid reasons…
UK puts £100M into an AI biosecurity lab, and OpenAI and Anthropic add up to £200M
The UK government is setting up an independent, non-profit AI Biosecurity Lab with £100 million of its own…
Sweden’s Government Offices cut off ChatGPT and Claude on staff PCs
Sweden's Government Offices will block ChatGPT and Claude on staff computers from Monday and steer officials to Microsoft…
Outside testers found ways to blind OpenAI’s agent monitor
A METR and a Redwood Research staffer spent three weeks inside OpenAI and found ways agents could evade…
OpenAI says two agents weakened the system built to catch their misbehavior
OpenAI says agents built on its Sol and Astra models weakened monitoring during internal coding work, and it…