Artificial Intelligence
Researchers at Gen Threat Labs showed an AI agent can be tricked by email content into running commands and deleting files. A vendor-sponsored article outlines layered defences, from skill vetting to runtime controls.
ISO/IEC 27090:2026 gives guidance on cybersecurity threats specific to AI systems. For organizations already running ISO/IEC 27001 and 42001, the most efficient path is to extend what they have rather than build new standalone structures.
A new comparison of AI-driven and human penetration testing finds each approach has distinct strengths, suggesting the two work best together rather than as substitutes.
Workers who paste company data into personal ChatGPT accounts can bypass enterprise safeguards; security vendor NordLayer outlines how IT teams can detect and prevent this kind of shadow AI use.
A phishing campaign uses fake AI assistant websites and browser-in-browser tricks to steal advertising account logins and MFA codes from agency staff and media buyers.
Security researcher Bruce Schneier argues media coverage of AI systems acting unexpectedly misuses terms like "hacking" and "going rogue," obscuring that responsibility lies with those who deployed the AI.
Research into infostealer malware logs found stolen AI account credentials linked to more than 80,000 corporate domains, exposing risks from unsanctioned "shadow AI" use to "LLMjacking" of paid accounts.
Google has released an interactive, open-access version of its AI & Economy ATLAS, letting anyone explore millions of data points on how AI is affecting economies worldwide.
An AI system called GPT-6 Astra reportedly broke a previously unsolved WWII Enigma message, building its own codebreaking tools along the way. Here is what happened and why it matters for cryptography and AI safety.
A US defence budget request seeks $30.3 million over five years for "Polygraph+", which would pair AI scoring with camera-based, contact-free sensing. Researchers warn the science behind lie detection remains weak.
A long Anthropic report on detected misuse of its Claude models, summarised into 117 findings, shows AI agents increasingly handling attack work while humans choose targets and review results.
Researchers have found a worm-like botnet called Carbonato that hijacks Docker servers left open to the internet and installs an AI agent to steal credentials and run commands on the attackers' behalf.
Jacob Coxon spent three years training frontier models at OpenAI and Anthropic. In a seven-post thread announcing his resignation, he argues both labs privately believe their technology could kill everyone within the decade — and are racing toward it anyway because neither trusts the other to stop.
Google's new certificate teaches practical, everyday AI skills, communication, research, data analysis, and no-code app building, aimed at closing a wide gap between what managers expect from AI and what workers have actually been trained on.
A practical template and reasoning for the policy every organization now needs: what staff can and cannot put into AI tools, and how to make the policy something people actually read.
As AI tools spread across organizations, governance policy, not just technical controls, determines whether adoption is safe, compliant, and trustworthy.
AI safety for employees means knowing what can go wrong when you use AI tools at work, misplaced trust in outputs, manipulation of the AI itself, and data exposure, and how to use them without creating risk for yourself or your employer.
Anthropic is embedding an invisible statistical watermark in Claude output, giving verification tools a way to flag AI-generated text and images without changing how the content looks or reads.
A mainstream browser now automatically flags images carrying C2PA provenance data, surfacing an AI-generated badge without requiring an extension or any technical know-how from the user.
An employee pastes a contract into a free AI tool to get a quick summary. It does not feel like a security incident. It might be one.