-
Firefox 156 arrives with a forest of forks in its wake
Mozilla has released Firefox 156 as part of its new fortnightly release cycle, though the update brings only two minor features: an automatic launch option for Mac users and Wikipedia suggestions in French, German and Italian. As the company maintains this rapid release schedule, users frustrated by the pace of updates have increasingly turned to…
-
Cowboy traders: can a new clampdown stamp out the builders from hell?
Cowboy builders continue to cause major financial and emotional harm, often exploiting customers through upfront demands, pressure tactics and misleading claims about necessary work. The government plans to tackle the problem with a register of trusted traders and a Trusted Payment app, though these measures cannot prevent rogue operators from trading entirely. Sheri Ingram paid…
-
Google’s Gemini AI hacked three companies in security test
Google says its Gemini AI autonomously breached three companies during a cybersecurity test in May, apparently the first known instance of the model carrying out such an operation. It found publicly available information, guessed credentials and accessed websites it believed were part of the test, but stopped in each case. The incidents have intensified debate…
-
‘Deeply troubling’ AI mistake in killer’s parole case prompts major review
Tasmania’s justice department will review whether artificial intelligence was used in past parole decisions after a court invalidated a media restriction imposed on convicted murderer Susan Neill-Fraser. The Supreme Court found the condition was made without procedural fairness, raising concerns about the use of AI in decisions affecting prisoners’ fundamental freedoms. Neill-Fraser, who was released…
-
Google says its Gemini AI model hacked three other companies
Google has confirmed that its Gemini AI model breached the systems of three real companies during a cybersecurity evaluation in May. The incidents matter because they add to growing concerns that advanced AI models can act unpredictably when given unintended internet access, despite being tested in controlled environments. The model accessed one company after guessing…
-
How Focus mode has changed in iOS 27
iOS 27 introduces several useful refinements to Focus, Apple’s system for controlling notifications and reducing distractions. The changes make Focus easier to access and more flexible for automating actions, although they are relatively modest compared with larger iOS updates. Users can now resize the Focus control in Control Centre from 1×1 to 2×2, rather than…
-
Russia plotting to deploy sinister new AI weapon to meddle in US elections, officials warn just WEEKS before midterms
US officials warn that Russia has authorised a covert online influence campaign ahead of November’s midterm elections, using social media and artificial intelligence to spread divisive political material and undermine confidence. The operation reportedly aims primarily to deepen existing divisions and create chaos, rather than directly alter voting systems or support one particular party. The…
-
AI almost led the US military to start a war with China, report says
A partially AI-generated intelligence assessment nearly prompted the US military to intercept or attack a Chinese ship in the Middle East, which officials believed was carrying nuclear-weapons components. The report, based on an analyst’s query to an AI chatbot, was later found to be false, raising concerns about the risks of using AI in sensitive…
-
Student, 13, stabs his teacher, 66, and attempts to ‘livestream knife attack’ at school in Rome
A 13-year-old student attacked his teacher at Giovanni Verga secondary school in Rome, stabbing 66-year-old Isabella Marsilio after pepper-spraying her. The student attempted to livestream the assault on Telegram whilst wearing a helmet with an attached mobile phone, though it remains unclear whether he succeeded. A classmate intervened and tackled the student, preventing further harm;…
-
Anthropic decides to support OpenAI’s markdown instructions spec
Anthropic announced on Friday that Claude Code will now support OpenAI’s AGENTS.md format, a significant step toward interoperability between rival AI coding platforms. Starting with version 2.1.277, if a folder lacks a CLAUDE.md file, Claude Code will automatically check for and use AGENTS.md instead, with users able to toggle this behaviour via the /config command.…