AI Safety stories
Gartner warns most AI projects may fail as enterprises struggle to track sensitive data that new tools and agents can access.
Enterprises will get tighter AI controls as Snowflake adds blocking policies, multi-party authorisation and new agentic tools at Summit.
Windows users can now run trillion-parameter AI models locally as NVIDIA targets enterprise developers with a deskside workstation and new security controls.
AWS customers can now run OpenAI's latest models in production without leaving Bedrock, with pricing and governance folded into existing commitments.
Software groups in chip design and healthcare are already using Nvidia's new agent tools to automate complex workflows with tighter security controls.
New silicon-level controls aim to curb unauthorised agent access and data exposure in enterprise AI storage, while keeping traffic fast.
Companies can now tie AI code-use risks to developer training, with Secure Code Warrior aiming to prove compliance at commit level.
The recognition bolsters confidence in BoodleBox as colleges and universities weigh transparent AI tools against concerns over governance and classroom use.
The deal is set to deepen research ties and speed up commercial use of quantum technologies, as Ottawa seeks trusted partners in a sensitive field.
Enterprises face uneven safeguards as TELUS Digital found no generative AI model was fully immune to attack in 620,000 tests.
Enterprises using AI tools may now face a tougher check on their defences as benchmark scores give way to real-world attack testing.
It targets operators where outages can threaten safety and continuity, as industrial and healthcare environments face faster-moving AI-driven attacks.
Regulated firms could gain tighter oversight and faster document workflows as Hyland adds agent controls, context tools and sector packages.
European businesses face fresh compliance risk after a study found all 12 leading AI models breached core checks under EU law in testing.
The tie-up aims to cut manual booking work and speed travel software delivery as AI chat tools reshape how trips are planned and sold.
The platform targets manual drug-development bottlenecks, from trial documents to safety case intake, as biopharma seeks faster compliant workflows.
Security, privacy and skills shortages are slowing Australian agencies, even as most weigh sovereign AI for defence and public health.
Regulated employers can now test their AI controls in minutes, as the Brisbane firm targets stricter data rules and workplace leak risks.
Accountants cannot yet use Claude to pull invoices or amend records, as Xero's new link is limited to read-only financial summaries.
Adoption has surged to 17.4 million users, even as most Australians remain uneasy about tech firms' data use and ad-funded answers.
